Skip Navigation


Age and Ageing Advance Access originally published online on December 6, 2007
Age and Ageing 2008 37(2):194-200; doi:10.1093/ageing/afm171
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
37/2/194    most recent
afm171v1
Right arrow E-letters: Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when E-letters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (5)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Ford, J.
Right arrow Articles by Dobson, A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Ford, J.
Right arrow Articles by Dobson, A.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Copyright © The Author 2007. Published by Oxford University Press on behalf of the British Geriatrics Society.

Self-rated health and a healthy lifestyle are the most important predictors of survival in elderly women

Jessica Ford, Melanie Spallek and Annette Dobson

School of Population Health, Faculty of Health Sciences, University of Queensland, Herston, QLD 4006, Australia

Address correspondence to: Melanie Spallek. Tel: +61 (0)7 334 64691; Fax: +61 (0)7 3365 5540. Email: m.spallek{at}sph.uq.edu.au


   Abstract
Top
 Abstract
Introduction
 Methods
 Results
 Discussion
 Key points
 References
 
Objective: to test the hypothesis that morbidity and health related behavioural factors are stronger than social factors as predictors of death among older women.

Methods: we used data from 12,422 participants in the Australian Longitudinal Study on Women's Health who were aged 70–75 in 1996. Proportional hazards models of survival up to 31 October 2005 were fitted separately for the whole cohort and those women who were initially in ‘good health’.

Results: among the whole cohort, 18.7% died during the follow_up period. The strongest predictor of death was ‘poor’ or ‘fair’ self-rated health (with 52.3% and 28.0%, respectively, of women in these categories dying). Among the women in ‘good health’ at baseline 11.5% died, with current cigarette smoking (hazard ratio HR = 2.19, 95% confidence interval (1.71, 2.81), physical inactivity (HR = 1.45 (1.17, 1.81)), and age (HR = 1.10 (1.04, 1.16) per year) as statistically significant predictors of death.

Discussion: among older women, current health and health related behaviours are stronger predictors than social factors of relatively early mortality. Adopting a healthier lifestyle, by doing more exercise and not smoking, is beneficial even in old age.

Keywords: self-rated health, survival, elderly women, predictors, elderly


   Introduction
Top
 Abstract
 Introduction
Methods
 Results
 Discussion
 Key points
 References
 
Life expectancy, particularly among older people, is determined by a complex mix of genetic and environmental factors and chance [1]. In younger age, genetic susceptibility, socio-demographic and behavioural factors are strongly related to mortality. But it has been argued that among older people, these predictors are less important, and death is best predicted by the accumulation of multiple morbidity and disability [2]. For example, empirical analysis has shown that the association between socio-demographic factors and mortality weakens when health status is taken into account [3]. Also socio-economic differentials in mortality appear to decrease in magnitude with increasing age [4].

These findings can be viewed from a life-course perspective. Many socio-demographic factors have their origins in early life; parental socio-economic status influences a person's educational and occupational aspirations and achievements, which in turn affect their own socio-economic status. These factors influence the establishment of lifestyle characteristics such as tobacco and alcohol consumption, levels of physical activity, weight maintenance and other health-related behaviours. In turn, these behavioural factors (together with genetic makeup, environmental exposures and chance) contribute to the accumulation of health problems. From this perspective one might hypothesize that by the time a person reaches old age their survival will be best predicted by the most proximal factors of morbidity and disability, which are a result of accumulating lifestyle and social factors.

For this paper, data from the older cohort of the Australian Longitudinal Study on Women's Health (ALSWH) were used to test the hypothesis and examine the relative importance of morbidity, health-related behavioural factors and social factors as predictors of death.


   Methods
Top
 Abstract
 Introduction
 Methods
Results
 Discussion
 Key points
 References
 
The ALSWH is a prospective study of factors affecting the health and well-being of three cohorts of women aged 18–23 years (‘younger cohort’), aged 45–50 years (‘mid-aged cohort’) and aged 70–75 years (‘older cohort’) at the time of Survey 1 in 1996. The study sample was selected randomly from the Medicare Australia database (which covers all citizens and permanent residents of Australia, including refugees and immigrants), with intentional over-representation of women living in rural and remote areas. Further details of the recruitment methods and response rates have been described elsewhere [5] and details of the study can be found at www.alswh.org.au.

The focus of this paper was the older cohort. Self-reported survey data were used and identification information was probabilistically matched to the National Death Index to identify deaths between study entry and 31 October 2005.

Measures
The baseline questionnaire covered health status, health-related behaviour, socio-demographic characteristics and social support. Several variables in each area were selected for this analysis based on previously reported associations with mortality in older adults.


Health status
The questionnaire included the Australian version of the SF-36 Health Profile, which assesses general physical and psychological health and well-being over the last 4 weeks [6]. There is a ‘General Health’ item that asks participants to rate their health, with possible responses ranging from ‘excellent’ to ‘poor’. This response is used as a global measure of self-rated health.

The mental health (MH) score from SF-36 is based on a weighted sum of five items, which include one or more items from each of the four major MH dimensions (anxiety, depression, loss of behavioural/emotional control, and psychological well-being) [7]. The criterion of MH ≤52 is used to indicate poor MH [8].

Co-morbidity has been shown to predict mortality [9]. A co-morbidity score was created as a weighted sum of the self-reported conditions found to be significantly predictive of mortality in the ALSWH. These were: stroke, diabetes and any cancer (excluding skin cancer) each with a weight of three; heart disease, diabetes or low iron, each with a weight of two; and osteoporosis with a weight of one. The weights were simplified from a Cox proportional hazards model. The co-morbidity score was then categorised into three groups: 0, 1–2, >2.


Health-related behavioral factors
Smoking status, level of physical activity, alcohol consumption and body mass index (BMI) have previously been shown to be associated with early mortality in the elderly and were selected as indicators of health [3, 10, 11]. Smoking status was categorised as ‘never smoker’, ‘ex-smoker’ or ‘current smoker’. Additionally, ex-smokers were classified into the following categories based on how long ago they gave up smoking: ‘quit within the last 5 years’, ‘quit 5–10 years ago’, ‘quit 11–20 years ago’, and ‘quit more than 20 years ago’. Responses about frequency of participation in ‘vigorous’ and ‘less vigorous’ activity were used to derive a physical activity score [12]. Categories of physical activity were labeled as ‘none’, ‘low’, ‘moderate’ and ‘high’. Alcohol consumption was categorised as: ‘non-drinker’, ‘rarely drinks alcohol’, ‘low-risk drinker’ (≤14 drinks per week) or ‘risky drinker’ (>14 drinks per week), adapted from Australian National Health and Medical Research Council [13] guidelines. BMI was calculated as weight (kg) divided by the square of height (m), then categorised as ‘underweight’ (BMI <18.5 kg/m2), ‘acceptable weight’ (BMI 18.5—<25 kg/m2), ‘overweight’ (BMI 25—<30 kg/m2), or ‘obese’ (BMI ≥30 kg/m2), according to the World Health Organization guidelines [14].


Socio-demographic factors
Age in years at the baseline survey in 1996, ranging from 70 to 75 years was included among the possible explanatory variables. Education level, based on highest qualification achieved, was categorised as: ‘no formal qualification’, ‘school certificate’ (school or higher school certificate), ‘trade’ (trade, certificate, diploma) and ‘university’ (completed a university degree). Women were asked how well they managed on their available income, with the following response options: ‘impossible or difficult all the time’, ‘difficult sometimes’, ‘not too bad’ and ‘easy’. Women were also asked about their housing situation with response options categorised as: ‘house’, ‘flat/unit/apartment’ or ‘other’. On the basis of an index of distance to the nearest urban centre, area of residence was classified as ‘urban’, ‘large rural centre’, ‘small rural centre’ and ‘other rural or remote area’. Country of birth was categorised as: ‘Australia’, ‘other English speaking’ and ‘other’.


Social support measures
The social support measures chosen were marital status, living arrangements, social interaction and whether there had been a decline in the health of their spouse or partner in the last 12 months. Marital status at baseline was categorised as: ‘single’, ‘partnered’ or ‘widowed’. Level of social interaction was calculated using a four-item subscale from the 11-item Duke Social Support Index [15]. Women were asked to report whether or not they had experienced a series of life events within the previous 12 months including a major decline in the health of their spouse or partner.

Analysis
The survival time was calculated as the number of days between the return of the baseline survey and the date of death or survival at 31 October 2005.

All analyses were performed using SAS statistical software version 9.1 [28]. Proportional hazard survival analyses were conducted for each potential explanatory variable (covariate) separately. The proportional hazard assumption for each potential explanatory variable to be used in the model was checked using the ASSESS PH statement in PROC PHREG procedure. Covariates with a P-value<0.1 from the univariate analyses were then included in a stepwise proportional hazards model using PROC TPHREG.

Owing to the overwhelming importance of poor self-rated health as a predictor of death, a second analysis focused on the group of women who were ‘healthy’ at the baseline survey. Thus women who rated their health as ‘fair’ or ‘poor’; those with BMI <18.5; women who reported having been told they had heart disease, stroke, diabetes or cancer (other than skin cancer); or those who indicated that they were ‘limited a lot’ in walking 100 m were excluded.


   Results
Top
 Abstract
 Introduction
 Methods
 Results
Discussion
 Key points
 References
 
By 31 October 2005, there had been 2,321 (18.7%) deaths among the 12,422 older women who participated in the baseline survey. Among those who had rated their health as ‘poor’, 52.3% had died, and among those who reported ‘fair’ health, 28.0% had died. After excluding women with poorer health there were 6,576 women (52.9% of the whole cohort) who were in good health at baseline. Of these ‘healthy’ women, 759 (11.5%) died during the follow-up period.

Women in the ‘healthy’ group had lower co-morbidity scores, higher BMI, higher MH scores, and were less likely to be sedentary than those in the whole cohort (results not shown here).

The factors statistically significantly associated with increased mortality in the whole cohort were: poorer self-rated health, low MH scores, higher co-morbidity score, being a current or ex-smoker, physical inactivity, being a non-drinker of alcohol, being underweight, older age, lower educational level, difficulty managing on available income, not living in a house, being single or widowed, living arrangements other than living with a spouse or partner and lower social interaction score. Several of these factors were not statistically significant for the ‘healthy’ women; these were mainly socio-demographic factors such as education, managing on their income, housing and social interactions, together with BMI (see Table 1). The proportional hazard assumption was met by all covariates for both cohorts. When all variables with P-value <0.1 from the univariate analyses were included in a multivariable stepwise proportional hazards model for the whole cohort, the variables that best predicted early death were: self-rated health of ‘poor’ or ‘fair’, physical inactivity, being a current smoker or an ex-smoker who had quit within the last 20 years, having a high co-morbidity score, being underweight, widowed and older aged, whilst being overweight was protective (see Table 2).


View this table:
[in this window]
[in a new window]

  		Table 1. Predictors of death among older women in the whole cohort and in the ‘healthy’ cohort: results from univariate proportional hazards survival analyses—number in each category (N), percent who died (%), hazard ratios (HR), and 95% confidence intervals (CI)

 


View this table:
[in this window]
[in a new window]

  		Table 2. Predictors of death among older women in the whole cohort and in the ‘healthy’ cohort: results from a stepwise proportional hazards survival analysis—hazard ratios (HR) and 95% confidence intervals (CI) for all factors that had P-values <0.1 in the univariate analyses

 
Among the women who were ‘healthy’ at baseline, the variables that best predicted death were current smoking, having low levels of physical activity, ‘good’ (compared to ‘excellent’) self-rated health and older age. The other variables predicting death which were significant from the univariate analyses (MH, alcohol consumption, country of birth and marital status) were not significant when the other factors were taken into account.


   Discussion
Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
Key points
 References
 
The baseline survey of the ALSWH was conducted in 1996 when women in the older cohort were aged 70–75 years and their average life expectancy was 14 years [16]. Thus, women who died in the first 9 years of the Study represent ‘early’ deaths at a life stage when a range of predictive factors might be expected to be important. However, the results presented here provide strong support for the hypothesis that for older women morbidity is the strongest predictor of mortality, followed by various health-related lifestyle factors. Social factors were not consistently significantly associated with mortality among the whole cohort or the healthy sub-cohort.

The hypothesis, derived from a life-course perspective, supports findings of other studies. Nybo, et al. [2] and Korten et al. [17] both found that social and lifestyle factors lose strength in predicting mortality, when physical health and cognitive performance were taken into account.

The US Cardiovascular Health Study found that among men and women aged over 65 years the strongest predictors of mortality were objective measures of sub-clinical disease and disease severity, with weight, smoking history and physical inactivity (together with gender and age) also being important [18]. The role of body weight in older people seems complex, with the possible J or U-shaped mortality curves. Diehr et al. [19] found higher mortality among older adults with very low BMI compared to others. The importance of lack of physical activity as a predictor of mortality in older adults has been shown by Stessman, et al. [20] although it was suggested that no additional benefit accrues from more than moderate physical activity. Current smoking is another well-established strong predictor of mortality [21]. The findings for both smoking and physical activity are important for public health policy and practice as they suggest that the health gains from quitting smoking and maintaining a moderate level of physical activity extend into old age.

Results of the present study are in agreement with the well-known widowhood effect, in which the risk of death is highest within the first months of bereavement [22].

Taking initial health status into account has been found to considerably weaken the association between mortality and socio-economic characteristics [3]. Indeed, there is substantial evidence of the value of self-rated health as a predictor of ‘early’ mortality among women [23, 24]. Self-rated health has even been suggested to be as good a predictor of mortality in women as objective measures of health [25].

There may be several explanations for the relatively weaker associations between social factors and mortality in the elderly, compared to the more proximal factors of morbidity and health behaviour. Social factors may not have been measured adequately (although in our study we used well-validated items and scales). Additionally, in some fairly homogeneous populations, lack of variation in social conditions may account for their failure to predict mortality [17]. The alternative explanation provided by the concept examined in this paper is that the social factors act predominantly as earlier determinants of morbidity and lifestyle in old age, so that there is little residual effect once morbidity and behavioural factors have been taken into account. Where conflicting results have been reported, often the three major types of predictors—morbidity/disability, health-related behavioural factors, and social factors—have not been simultaneously measured or taken into account [11, 26].

One limitation of the present study is that the data are self-reported. While this is advantageous for variables like self-reported health, there could be some bias in the reported health behaviour. Unfortunately, cognitive function was not measured at the baseline survey of the ALSWH so it could not be included in the analysis, although it has been found by others to be a strong predictor of mortality in the elderly [17]. Another limitation is that the ALSWH provides data only for women—mortality for men may be associated with different factors [17].

The strengths of the study are that it is based on a randomly selected community sample; it is large, providing statistical power to examine the effects of many variables. Identification of deaths is possible through linkage to the National Death Index, where the true positive rate for the ALSWH was previously found to be 95% [27]. Furthermore, multiple measures of morbidity and disability, health-related behavior, socio-demographic characteristics of participants and social support are available.

In summary, data from the ALSWH provide support for the life-course perspective on factors associated with mortality in older adults. In this framework, one can expect that: morbidity and disability being most proximal to death provide the strongest predictors; lifestyle factors that are major risk factors for death among middle-aged people are weaker predictors among the elderly; and differences in social factors are less predictive of mortality among people who survive to older ages.


   Key points
Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Key points
References
 

  • Strongest predictors of early mortality among older women are current health and health-related behaviours.
  • Differences in social factors are less predictive of mortality among women who survive to older ages.
  • Adopting a healthier lifestyle, by doing more exercise and not smoking, is beneficial even in old age.


   Conflicts of interest
 
There are neither financial nor dual commitments that represent potential conflicts of interest.


   Acknowledgements
 
The Australian Longitudinal Study on Women's Health, which was conceived and developed by groups of interdisciplinary researchers at the universities of Newcastle and Queensland, is funded by the Australian Government Department of Health and Ageing. We thank all participants for their valuable contribution to this project. Details about the Australian Longitudinal Study on Women's Health can be found at www.alswh.org.au.


   References
Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Key points
 References
 

  1. Menotti A, Kromhout D, Nissinen A, et al. Short-term all-cause mortality and its determinants in elderly male populations in Finland, the Netherlands, and Italy: The FINE study. Prev Med (1996) 25:319–26.[CrossRef][Web of Science][Medline]
  2. Nybo H, Petersen HC, Gaist D, et al. Predictors of mortality in 2,249 nonagenarians–The Danish 1905–cohort survey. J Am Geriatr Soc (2003) 51:1365–73.[CrossRef][Web of Science][Medline]
  3. Mete C. Predictors of elderly mortality: health status, socioeconomic characteristics and social determinants of health. Health Econ (2005) 14:135–48.[CrossRef][Web of Science][Medline]
  4. Mishra GD, Ball K, Dobson AJ, et al. Do socioeconomic gradients in women's health widen over time and with age? Soc Sci Med (2004) 58:1585–95.[CrossRef][Web of Science][Medline]
  5. Lee C, Dobson AJ, Brown W, et al. Cohort profile: The Australian Longitudinal Study on Women's Health. Int J Epidemiol (2005) 34:987–91.[Free Full Text]
  6. McCallum J. The SF-36 in an Australian sample: validating a new, generic health status measure. Aust J Public Health (1995) 19:160–6.[Web of Science][Medline]
  7. Ware JE, Kosinski M, Keller SD. SF-36 Physical and Mental Health Summary Scales: A User's Manual (1994) 12. Boston, MA: The Health Institute, New England Medical Center.
  8. Holmes WC. A short, psychiatric, case-finding measure for HIV seropositive outpatients: performance characteristics of the 5-item mental health subscale of the SF-20 in a male, seropositive sample. Med Care (1998) 36:237–43.[CrossRef][Web of Science][Medline]
  9. Byles J, D'Este C, Parkinson L, et al. Single index of morbidity did not predict multiple outcomes. J Clin Epidemiol (2005) 58:991–1005.[CrossRef][Web of Science][Medline]
  10. Ahmad R, Bath PA. Identification of risk factors for 15 year mortality among community-dwelling older people using Cox regression and a genetic algorithm. J Gerontol Ser A 2005: Biol Sci Med Sci (2005) 60:1052–8.
  11. Bath PA, Morgan K. Customary physical activity and physical health outcomes in later life. Age Ageing (1998) 27-S3:29–34.
  12. Brown WJ, Mishra G, Lee C, et al. Leisure time physical activity in Australian women: relationship with well-being and symptoms. Res Q Exerc Sport (2000) 71:206–16.[Web of Science][Medline]
  13. National Health and Medical Research Council. Australian Alcohol Guidelines: Health Risks and Benefits. Endorsed October 2002. (2001) Canberra, Australia.
  14. WHO Consultation on Obesity. Obesity: Preventing and Managing the Global Epidemic: Report of a WHO Consultation. (1999) WHO Technical Report Series 894, Geneva.
  15. Goodger B, Byles J, Higginbotham N, et al. Assessment of a short scale to measure social support among older people. Aust N Z J Public Health (1999) 23:260–5.[Web of Science][Medline]
  16. Australian Prudential Regulatory Authority. Australian Life Tables 1995-97 (1999) Canberra: Australian Government Publishing Service.
  17. Korten AE, Jorm AF, Jiao Z, et al. Health, cognitive, and psychosocial factors as predictors of mortality in an elderly community sample. J Epidemiol Community Health (1999) 53:83–8.[Abstract]
  18. Burke GL, Arnold AM, Bild DE, et al. Factors associated with health aging: The Cardiovascular Health Study. J Am Geriatr Soc (2001) 49:254–62.[CrossRef][Web of Science][Medline]
  19. Diehr P, Bild DE, Harris TB, et al. Body mass index and mortality in nonsmoking older adults: The Cardiovascular Health Study. Am J Public Health (1998) 88:623–9.[Abstract/Free Full Text]
  20. Stessman J, Maaravi Y, Hammerman-Rozenberg R, et al. The effects of physical activity on mortality in the Jerusalem 70 year-olds longitudinal study. J Am Geriatr Soc (2000) 48:499–504.[Web of Science][Medline]
  21. Peto R, Lopez AD, Boreham J, et al. Mortality from Smoking in Developed Countries, 1950–2000: Indirect Estimates from National Vital Statistics (1994) Oxford: Oxford University Press.
  22. Elwert F, Christakis NA. Widowhood and race. Am Sociol Rev (2006) 71:16–41.[Web of Science]
  23. McCallum J, Shadbolt B, Wang D. Self-rated health and survival: A 7-year follow-up study of Australian elderly. Am J Public Health (1994) 84:1100–5.[Abstract/Free Full Text]
  24. Onawola RST, LaVeist TA. Subjective health status as a determinant of mortality among African American elders. J Natl Med Assoc (1998) 90:754–8.[Medline]
  25. Mossey JM, Shapiro E. Self-rated health: A predictor of mortality among the elderly. Am J Public Health (1982) 72:800–8.[Abstract/Free Full Text]
  26. Yu SM, Chie WC, Chen TH-H, et al. Findings of periodic health examination and risk of mortality in a cohort of elderly people in Taiwan. Asia Pac Fam Med (2004) 3:7–12.
  27. Powers J, Ball J, Adamson L, et al. Effectiveness of the national death index for establishing the vital status of older women in the Australian Longitudinal Study on Women's Health. Aust N Z J Public Health (2000) 24:526–8.[Web of Science][Medline]
  28. SAS. SAS/STAT Users Guide (1999) Cary, NC: SAS Inst.
Received 29 April 2007; accepted in revised form 17 September 2007.


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 Age AgeingHome page
T. Sulander, O. Rahkonen, O. Nummela, and A. Uutela
Ten year trends in health inequalities among older people, 1993-2003
Age Ageing, September 1, 2009; 38(5): 613 - 617.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
37/2/194    most recent
afm171v1
Right arrow E-letters: Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when E-letters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (5)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Ford, J.
Right arrow Articles by Dobson, A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Ford, J.
Right arrow Articles by Dobson, A.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?