Abstract
Aim
To describe the Dynamic Analyses to Optimise Ageing (DYNOPTA) project and illustrate its contributions to understanding ageing through innovative methodology, and investigations on outcomes based on the project themes. DYNOPTA provides a platform and technical expertise that may be used to combine other national and international datasets.
Method
The DYNOPTA project has pooled and harmonized data from nine Australian longitudinal studies to create the largest available longitudinal dataset (N=50652) on ageing in Australia.
Results
A range of findings have resulted from the study to date, including methodological advances, prevalence rates of disease and disability, and mapping trajectories of ageing with and without increasing morbidity. DYNOPTA also forms the basis of a microsimulation model that will provide projections of future costs of disease and disability for the baby boomer cohort.
Conclusion
DYNOPTA contributes significantly to the Australian evidence-base on ageing to inform key social and health policy domains.
Keywords: cognitive decline, data pooling, disability, longitudinal study, mental health
INTRODUCTION
With the unprecedented change in population ageing occurring in Australia(1) and internationally(2), there is a strong need for governments, industry and the general public to become better informed about human ageing from biological, social and psychological perspectives. However, without adequate information, implicit views about ageing, rather than research, may influence policy decisions relating to older adults.
The discipline of Gerontology was formed in the 1960s, spurning the instigation of early longitudinal studies focusing on older adults. Since then life expectancies have increased substantially, and the early studies, whilst groundbreaking, often do not contain many participants who lived past 85 years of age. Therefore, there is an increasing need for information on ageing of adults aged 80 and older. Moreover, there is also a great need for an understanding of individual ageing as a developmental process occurring over time, contextualized in socio-political, environmental and historical contexts. Clearly, initiating such a study now would require decades to pass before sufficient information on ageing and change was available. The Dynamic Analyses to Optimise Ageing Project (DYNOPTA) pooled dataset provides the largest available evidence base on ageing of older Australian adults The longitudinal data allow for the analysis of prevalence, incidence and change in major causes of disability among persons as they age, The studies enables comparison across successive cohorts, providing evidence of current and projected disability burdens.
THE DYNOPTA PROJECT
The broad aims of DYNOPTA reflect the vision of the Prime Minister’s Science Engineering and Innovation Council (PMSEIC) report(3) to identify effective pathways to compressing morbidity and optimising ageing. Since commencing in 2007, the DYNOPTA project has produced a pooled dataset comprising information from nine Australian Longitudinal Studies of Ageing. Data were harmonised from the contributing studies (described in Table 1) to create an entirely new and unique dataset. Where possible, variables were harmonised to enable comparison with Australian benchmarks. For example, the physical activity data have been harmonised to derive measures reflecting the recommended level of physical activity per week(4) and the alcohol consumption data have been harmonised to provide classifications in accordance with those endorsed by the National Health and Medical Research Council(5).
Table 1.
Characteristics of Cohort studies Contributing to DYNOPTA
| Study | Study Acronym | Location | wave | year | N | age |
|---|---|---|---|---|---|---|
| Australian Longitudinal Study of Ageing Focus: Multidisciplinary, health ageing |
ALSA | Adelaide | 1 | 1992-93 | 2087 | 65-103 |
| 2 | 1993-94 | 1779 | 65-104 | |||
| 3 | 1994-95 | 1679 | 66-105 | |||
| 4 | 1995-96 | 1504 | 68-106 | |||
| 5 | 1998 | 1171 | 70-100 | |||
| 6 | 2000-01 | 791 | 72-101 | |||
|
| ||||||
| Australian Longitudinal Study of Women’s Health Focus: Women’s health |
ALSWH-mid cohort | National | 1 | 1996 | 12716 | 45-51 |
| 2 | 1998 | 12338 | 47-53 | |||
| 3 | 2001 | 11200 | 50-56 | |||
| 4 | 2004 | 10903 | 53-59 | |||
|
| ||||||
| Australian Longitudinal Study of Women’s Health Focus: Women’s health |
ALSWH-old cohort | National | 1 | 1996 | 12432 | 68-76 |
| 2 | 1999 | 10434 | 71-79 | |||
| 3 | 2002 | 8646 | 74-82 | |||
| 4 | 2005 | 7158 | 77-85 | |||
|
| ||||||
| Australian Diabetes and Obesity Survey Focus: diabetes, obesity |
AusDiab | National | 1 | 1999-00 | 7296 | 45-95 |
| 2 | 2004-05 | 4380 | 49-93 | |||
|
| ||||||
| Blue Mountains Eye Study Focus: vision and hearing |
BMES | Blue Mountains, NSW | 1 | 1992-93 | 3654 | 45-97 |
| 2 | 1997-99 | 2334 | 54-98 | |||
| 3 | 2002-04 | 1952 | 60-97 | |||
|
| ||||||
| Canberra Longitudinal Study Focus: dementia, depression |
CLS | Canberra, ACT and Queanbeyan, NSW | 1 | 1990-91 | 1043 | 70-100 |
| 2 | 1994-95 | 652 | 74-102 | |||
| 3 | 1998 | 382 | 78-101 | |||
| 4 | 2002 | 214 | 82-102 | |||
|
| ||||||
| Household, Income and Labour Dynamics of Australia Focus: economics, productivity |
HILDA | National | 1 | 2001 | 6164 | 45-90 |
|
| ||||||
| 2 | 2002 | 5454 | 45-90 | |||
|
| ||||||
| 3 | 2003 | 5089 | 46-90 | |||
|
| ||||||
| 4 | 2004 | 4769 | 47-90 | |||
|
| ||||||
| 5 | 2005 | 4658 | 48-90 | |||
| 6 | 2006 | |||||
|
| ||||||
| Melbourne Longitudinal Study Healthy Ageing Focus: healthy ageing, mobility |
MELSHA | Melbourne | 1 | 1994 | 1000 | 66-94 |
| 3 | 1996 | 796 | 68-98 | |||
| 5 | 1998 | 649 | 70-98 | |||
| 7 | 2000 | 522 | 72-100 | |||
| 9 | 2002 | 394 | 74-102 | |||
|
| ||||||
| Personality and Total Health through life Focus: mental health, cognition |
PATH | Canberra/Queanbeyan | 1 | 2001-02 | 2550 | 60-66 |
| 2 | 2005-06 | 2222 | 64-70 | |||
|
| ||||||
| Sydney Older Person’s study Focus: neurodegeneration |
SOPS | Sydney | 1 | 1991-93 | 647 | 75-99 |
| 2 | 1994-96 | 462 | 78-100 | |||
| 3 | 1996-97 | 379 | 78-99 | |||
| 4 | 1997-98 | 308 | 80-101 | |||
The DYNOPTA research program has four key foci, including dementia and cognition, mental health, sensory impairment, and mobility/activity limitations. Mortality is also included as a key outcome and other domains are studied as either risk or protective factors, or impacts of, the key foci. These include social networks, shared risk factors for chronic disease such as smoking, alcohol consumption, bodyweight, and physical activity, medical conditions, self-rated health and others. An interdisciplinary, life-course approach to ageing underpins the DYNOPTA research program (6, 7).
The DYNOPTA sample of men and women has been previously described (8). Briefly, the sample comprises 50,652 baseline participants (wave 1 of each study between 1990 and 2001). Of these, 39,085 (77.2%) were female, reflecting inclusion of the all-female Australian Longitudinal Study of Women’s Health (9) (particularly for the 45-54 and 65-74 age groups), and women’s greater longevity. Selected sample characteristics are shown in Table 2 and report frequency of participants’ characteristics as a proportion of the baseline sample. There was a higher likelihood of drop out in older age groups, overall 19.1% of the baseline sample did not participate in wave 2 due to either mortality or attrition. The majority of participants (61%) did not report obtaining qualifications after leaving secondary school. Only 11.2% of the sample were tertiary qualified, although the proportion of participants with post-secondary qualifications increased for younger age groups. There was an even distribution of career occupations, 30.3% reported career occupation to be managerial or professional, 28.7% clerical and associate professional, 8.3% tradespersons, and 32.7% worked in sales, service, production, transport or unskilled labourer roles. Of those with available data, 73.4% of the baseline sample were born in Australia and 88% reported English to be their native or preferred language.
Table 2.
Unweighted Sample Profile for Socio-demographic Variables by 10 Year Age-groups Baseline (N = 50,652).
| TOTAL | 45-54 | 55-64 | 65-74 | 75-84 | 85-94 | 95+ | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| n | (%) | n | (%) | n | (%) | n | (%) | n | (%) | n | (%) | n | (%) | |
| Gender | ||||||||||||||
| Male | 11567 | (22.8) | 2719 | (5.4) | 3459 | (6.8) | 2820 | (5.6) | 2062 | (4.1) | 492 | (1.0) | 15 | (0.0) |
| Female | 39085 | (77.2) | 16784 | (33.1) | 3674 | (7.3) | 15059 | (29.7) | 2919 | (5.8) | 591 | (1.2) | 58 | (0.1) |
| Partner Status | ||||||||||||||
| Married or defacto | 35199 | (69.9) | 15929 | (31.6) | 5492 | (10.9) | 10875 | (21.6) | 2556 | (5.1) | 339 | (0.7) | 8 | (0.0) |
| Divorced or separated | 4627 | (9.2) | 2350 | (4.7) | 891 | (1.8) | 1132 | (2.2) | 236 | (0.5) | 18 | (0.0) | 0 | (0.0) |
| Widowed | 8547 | (17.0) | 364 | (0.7) | 456 | (0.9) | 5057 | (10.0) | 1948 | (3.9) | 665 | (1.3) | 57 | (0.1) |
| Never married | 1959 | (3.9) | 783 | (1.6) | 286 | (0.6) | 601 | (1.2) | 224 | (0.4) | 58 | (0.1) | 7 | (0.0) |
| Highest Educational Attainment | ||||||||||||||
| Secondary school | 29860 | (61.0) | 11069 | (22.6) | 2584 | (5.3) | 12511 | (25.6) | 3065 | (6.3) | 595 | (1.2) | 36 | (0.1) |
| Non-tertiary | 13611 | (27.8) | 5199 | (10.6) | 3140 | (6.4) | 3632 | (7.4) | 1358 | (2.8) | 272 | (0.6) | 10 | (0.0) |
| Tertiary | 5484 | (11.2) | 2996 | (6.1) | 1249 | (2.6) | 922 | (1.9) | 256 | (0.5) | 58 | (0.1) | 3 | (0.0) |
| Career Occupation | ||||||||||||||
| Managers and Professionals | 12591 | (30.3) | 5828 | (14.0) | 2426 | (5.8) | 2987 | (7.2) | 1098 | (2.6) | 239 | (0.6) | 13 | (0.0) |
| Clerical and Associate professional | 11924 | (28.7) | 5366 | (12.9) | 1371 | (3.3) | 4205 | (10.1) | 827 | (2.0) | 150 | (0.4) | 5 | (0.0) |
| Tradespersons | 3444 | (8.3) | 1014 | (2.4) | 590 | (1.4) | 1140 | (2.7) | 569 | (1.4) | 129 | (0.3) | 2 | (0.0) |
| Sales, Service, Transport and Labourers | 13574 | (32.7) | 5666 | (13.6) | 2111 | (5.1) | 4271 | (10.3) | 1248 | (3.0) | 262 | (0.6) | 16 | (0.0) |
| Main Source of Income | ||||||||||||||
| Wages or salary | 4222 | (22.6) | 3033 | (16.2) | 1109 | (5.9) | 73 | (0.4) | 7 | (0.0) | 0 | (0.0) | 0 | (0.0) |
| Superannuation or investments | 1860 | (9.9) | 141 | (0.8) | 596 | (3.2) | 696 | (3.7) | 349 | (1.9) | 78 | (0.4) | 0 | (0.0) |
| Business or partnership | 1361 | (7.3) | 784 | (4.2) | 419 | (2.2) | 122 | (0.7) | 30 | (0.2) | 6 | (0.0) | 0 | (0.0) |
| Pension or government allowance | 8170 | (43.7) | 677 | (3.6) | 1481 | (7.9) | 2998 | (16.0) | 2471 | (13.2) | 523 | (2.8) | 20 | (0.1) |
| Multiple sources | 3092 | (16.5) | 569 | (3.0) | 383 | (2.0) | 1052 | (5.6) | 818 | (4.4) | 263 | (1.4) | 7 | (0.0) |
| Country of Birth | ||||||||||||||
| Born in Australia | 36121 | (73.4) | 14403 | (29.3) | 4639 | (9.4) | 12748 | (25.9) | 3567 | (7.2) | 720 | (1.5) | 44 | (0.1) |
| Not born in Australia | 13110 | (26.6) | 4926 | (10.0) | 2191 | (4.5) | 4314 | (8.8) | 1333 | (2.7) | 331 | (0.7) | 15 | (0.0) |
| First, Native or Preferred Language | ||||||||||||||
| English | 10950 | (88.0) | 2071 | (16.6) | 3384 | (27.2) | 2273 | (18.3) | 2431 | (19.5) | 738 | (5.9) | 53 | (0.4) |
| Other | 1500 | (12.0) | 346 | (2.8) | 552 | (4.4) | 337 | (2.7) | 216 | (1.7) | 48 | (0.4) | 1 | (0.0) |
| Accommodation | ||||||||||||||
| Community living | 44686 | (97.1) | 18158 | (39.4) | 4711 | (10.2) | 16202 | (35.2) | 4694 | (10.2) | 899 | (2.0) | 22 | (0.0) |
| Institution | 358 | (0.8) | 28 | (0.1) | 24 | (0.1) | 35 | (0.1) | 98 | (0.2) | 136 | (0.3) | 37 | (0.1) |
| Other | 988 | (2.1) | 461 | (1.0) | 105 | (0.2) | 358 | (0.8) | 58 | (0.1) | 6 | (0.0) | 0 | (0.0) |
| Attrition | ||||||||||||||
| Participation in wave 2 | 40971 | (80.9) | 16716 | (33.0) | 5662 | (11.2) | 14493 | (28.6) | 3470 | (6.9) | 609 | (1.2) | 21 | (0.0) |
| Non participation in wave 2 | 9681 | (19.1) | 2787 | (5.5) | 1471 | (2.9) | 3386 | (6.7) | 1511 | (3.0) | 474 | (0.9) | 52 | (0.1) |
Note: percentage (%) of whole sample
DYNOPTA’s sample size and broad coverage of geography, time and age enables topics to be studied with greater breadth than possible in any single study. This allows for the compilation of normative data on indicators of ageing and disability in the four outcome areas. Importantly, DYNOPTA has a large enough sample of adults aged over 85 to enable the in-depth analysis of this age group to be conducted that has hitherto not been possible within Australia. In the following sections we demonstrate the value of the DYNOPTA project in developing a unique platform for innovative research in domains that are high priority for building the evidence based for public policy
METHODOLOGICAL CONTRIBUTIONS
Development of the study dataset, infrastructure and methods
DYNOPTA has led to the development of research methods and infrastructure that provide a platform for a far broader research program involving collaborative analyses between large scale studies both nationally and internationally. Briefly, in creating the pooled dataset, the DYNOPTA working party spent two years working through the optimal methods of harmonizing each variable in the final dataset which now includes over 450 variables. In some instances this was straightforward as the coding of variables was similar. In other instances, entirely new variables were created using statistical techniques (e.g. latent variable modeling, by fiat harmonization) or with experts who were consulted to aid in harmonizing variables where required (e.g. physical activity). Descriptive statistics from a selection of variables from the dataset are shown in Table 3. Examples of the contributions in methodologies are described below and are already being applied to cross-national analyses including DYNOPTA and studies from the U.S.A., Japan and Korea.
Table 3.
Baseline frequency counts (n) and percentage (%) within ten year age groups for a selection of functional, health and lifestyle variables in DYNOPTA, and the number of contributing studies for each variable.
| TOTAL | 45-54 | 55-64 | 65-74 | 75-84 | 85-94 | 95+ | Studies | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| n | (%) | n | (%) | n | (%) | n | (%) | n | (%) | n | (%) | n | (%) | n | |
| Cognitive Impairment | |||||||||||||||
| MMSE < 24 | 593 | (7.0) | 0 | (0.0) | 53 | (1.8) | 73 | (3.2) | 277 | (11.4) | 164 | (23.0) | 26 | (56.5) | 6 |
| Sensory Impairment | |||||||||||||||
| Mild Hearing Loss | 1,160 | (32.4) | 0 | (0.0) | 69 | (11.1) | 464 | (33.2) | 501 | (42.5) | 123 | (34.4) | 3 | (21.4) | 2 |
| Moderate to severe Hearing Loss | 700 | (19.6) | 0 | (0.0) | 26 | (4.2) | 154 | (11.0) | 318 | (27.0) | 192 | (53.6) | 10 | (71.4) | 2 |
| Vision Loss | 3,337 | (17.9) | 12 | (2.4) | 166 | (3.7) | 650 | (9.8) | 1,498 | (29.3) | 953 | (53.6) | 58 | (74.4) | 6 |
| Medical Conditions*,a | |||||||||||||||
| Arthritis | 3,105 | (43.4) | 799 | (35.6) | 1,034 | (51.1) | 956 | (43.8) | 305 | (46.4) | 5 | ||||
| Diabetes | 516 | (7.2) | 169 | (7.5) | 135 | (6.7) | 171 | (7.8) | 36 | (5.5) | 5 | ||||
| Hypertension | 1,731 | (33.9) | 908 | (41.3) | 320 | (32.5) | 395 | (28.2) | 104 | (20.8) | 3 | ||||
| Self Rated Health | |||||||||||||||
| Excellent-Good | 35,305 | (79.9) | 16,375 | (87.6) | 4,844 | (82.9) | 11,373 | (73.1) | 2257 | (67.5) | 431 | (59.8) | 25 | (71.4) | 7 |
| Fair | 7,394 | (16.7) | 1,971 | (10.5) | 805 | (13.8) | 3,554 | (22.8) | 845 | (25.3) | 212 | (29.4) | 7 | (20.0) | 7 |
| Poor | 1,516 | (3.4) | 358 | (1.9) | 197 | (3.4) | 638 | (4.1) | 242 | (7.2) | 78 | (10.8) | 3 | (8.6) | 7 |
| Body Mass Indexa | |||||||||||||||
| Under-weight | 251 | (1.6) | 41 | (1.2) | 57 | (1.1) | 69 | (1.7) | 67 | (2.9) | 16 | (3.3) | 5 | ||
| Normal | 5,630 | (36.2) | 1,156 | (34.5) | 1,803 | (34.7) | 1,430 | (34.2) | 975 | (41.8) | 258 | (52.4) | 5 | ||
| Over-weight | 6,504 | (41.8) | 1,377 | (41.1) | 2,143 | (41.3) | 1,826 | (43.7) | 975 | (41.8) | 180 | (36.6) | 5 | ||
| Obese | 3,183 | (20.5) | 777 | (23.2) | 1,192 | (23.0) | 858 | (20.5) | 317 | (13.6) | 38 | (7.7) | 5 | ||
| Smoking | |||||||||||||||
| Never | 26,359 | (54.4) | 9,962 | (52.9) | 3,530 | (50.6) | 9,765 | (57.7) | 2,499 | (52.8) | 573 | (59.1) | 30 | (69.8) | 9 |
| Former | 15,858 | (32.7) | 5,436 | (28.9) | 2,479 | (35.5) | 5,705 | (33.7) | 1,880 | (39.7) | 346 | (35.7) | 12 | (27.9) | 9 |
| Current | 6,281 | (13.0) | 3,436 | (18.2) | 972 | (13.9) | 1,466 | (8.7) | 355 | (7.5) | 51 | (5.3) | 1 | (2.3) | 9 |
| Alcohol Long Term Risk | |||||||||||||||
| non-drinker | 9,931 | (21.5) | 2,520 | (13.4) | 742 | (11.3) | 5,027 | (30.8) | 1,273 | (33.8) | 353 | (45.6) | 16 | (61.5) | 8 |
| low risk | 33,486 | (72.4) | 15,130 | (80.4) | 5,058 | (77.3) | 10,521 | (64.6) | 2,363 | (62.8) | 404 | (52.1) | 10 | (38.5) | 8 |
| risky | 1,905 | (4.1) | 846 | (4.5) | 417 | (6.4) | 542 | (3.3) | 86 | (2.3) | 14 | (1.8) | 0 | (0.0) | 8 |
| high risk | 902 | (2.0) | 324 | (1.7) | 324 | (5.0) | 209 | (1.3) | 41 | (1.1) | 4 | (0.5) | 0 | (0.0) | 8 |
| Falls in past 12 months | |||||||||||||||
| none | 4364 | (69.3) | 361 | (82.1) | 859 | (79.3) | 1,357 | (69.0) | 1,412 | (66.3) | 359 | (55.4) | 16 | (61.5) | 5 |
| one | 1,195 | (19.0) | 57 | (13.0) | 154 | (14.2) | 396 | (20.1) | 434 | (20.4) | 147 | (22.7) | 7 | (26.9) | 5 |
| two or more | 734 | (11.7) | 22 | (5.0) | 70 | (6.5) | 213 | (10.8) | 284 | (13.3) | 142 | (21.9) | 3 | (11.5) | 5 |
| Walking 1kma | |||||||||||||||
| Not limited at all | 26,561 | (64.2) | 14,610 | (78.8) | 4,019 | (69.3) | 7,030 | (48.0) | 846 | (38.9) | 56 | (24.0) | 6 | ||
| Limited a little | 8,858 | (21.4) | 2,843 | (15.3) | 1,168 | (20.1) | 4,137 | (28.3) | 645 | (29.7) | 65 | (27.9) | 6 | ||
| Limited a lot | 5,971 | (14.4) | 1,088 | (5.9) | 611 | (10.5) | 3,477 | (23.7) | 682 | (31.4) | 112 | (48.1) | 6 | ||
Data unavailable for 45-54 at baseline of all studies;
cell sizes for 95+ too small to be reliable.
Note: Percentage (%) refers to number of individuals at baseline in that group with that condition.
Medical conditions are obtained by self-report; Vision loss is defined by presenting distance visual acuity greater than 0.3 logMAR in the better eye; Mild hearing loss is defined by averaged pure tone thresholds greater than 25 dB in the better ear; Moderate to severe hearing loss is defined by averaged pure tone thresholds greater than 40 dB in the better ear.
Novel approaches to imputation of study censored data
To maximize the investment and utility of combining datasets, the DYNOPTA project has developed methods of imputing missing data in order to allow whole datasets to be included in research projects. For example, one of the contributing studies did not obtain data on secondary schooling or low education attainment. The study in question was distinct in the DYNOPTA sample as it was the only contributing study that sampled a semi-urban population that was not a capital city and included data on cognitive and sensory functioning assessed via clinical interview. Because low education is known to be strongly associated with poor health outcomes in late life and this study’s specific sub population was unique in the DYNOPTA dataset, a binary indicator of age left school was imputed using novel logistic regression techniques(10).
Novel developments in handling missing data on multi-item scales
Even where contributing studies provided data on key variables, there were still significant issues relating to the handling of missing data in the harmonization process. For example, the Mini-Mental State Examination (MMSE) was a key cognitive measure available in 6 of the 9 contributing studies. Differences in how each contributing study computed total scores with missing item-level data required that DYNOPTA harmonization computed total scores from the raw MMSE data. Analysis of the DYNOPTA dataset has indicated that non-response on MMSE items is mostly unrelated to the likelihood of clinical diagnosis(11). Therefore adapting MI by computing an average of ‘n’ imputed datasets, we showed a significant increase in MMSE scores for those participants with missing data. Further, using a random sample of complete cases, indicated MI as an appropriate technique to resolve missing item level data when computing total MMSE scores (for a full description see (11). As in our analyses, MI has been demonstrated to be preferable to other techniques for dealing with missing data, including computing a pro-rata total score, mean-item substitution, and complete-case analysis techniques (12). Given the widespread use of the MMSE in ageing research, this work has broad application and utility to researchers in the field.
Novel developments in survey statistics
The complex structure of the DYNOPTA data has raised many statistical issues, one of which is how to determine an appropriate weight for each person in the DYNOPTA dataset so that the data from the contributing studies can be combined to produce results representative of the Australian population. These weights can then be used in the estimation and analysis of the prevalence and association between risk factors, disease, physical activity and mental health measures. One challenge for developing sample weights was the overlap in geographical coverage of some studies. Hence, a weight adjustment method has been developed which can be applied to the baseline weights for each study. The original baseline weights were used if already available for a study, and study weights were developed for wave 1 if they were not. The wave 1 weights adjusted the sample in each study to the specific population from which it was drawn to account for the probability of selection of each individual. The study weights were then calibrated to the Estimated Resident Population for the relevant year, gender, age-group and geographical area for each study.
To enable analysis of a pooled data set consisting of a combination of studies, final weights were developed which combined the study specific weights according to their contributing sample sizes to the pooled dataset, for each age-group by gender cell within each geographic region. The estimated standard errors for the prevalence estimates take into account the use of the final combined weights and the complex survey design for each study.
CONTRIBUTIONS TO KNOWLEDGE ABOUT AGEING IN AUSTRALIA
Identification of cohort effects in health behaviour - capitalizing on the different ages and dates of contributing samples
Because of the unique timing of the various contributing DYNOPTA studies, we are able to investigate possible cohort differences in change. In harmonizing the alcohol and smoking variables between DYNOPTA studies, two different cohort patterns of Australian alcohol and smoking consumption have been identified(13) which could be used to describe changes in the national consumption of alcohol and smoking. The results indicated that prevalence of high-risk alcohol consumption and the proportions of current smokers, had significantly declined for the cohort aged 45+ between 1990 and 1994 compared with the cohort who were the same age between 1996 and 2002. In relation to this finding, the authors concluded that government interventions, in the form of advertising campaigns, bans on smoking in public venues, and harsher penalties for alcohol-related crime, appeared to have had an impact on national alcohol and smoking consumption behavior.
While Australia is well served by national surveys of physical activity, sampling is restricted to those aged 75 and younger (14). However, in DYNOPTA our age range extends to 106 years allowing us to examine physical activity across a braoder range. Analyses have shown that men were more likely than women to ‘walk every day or more’. The prevalence of walking ‘everyday or more’ was relatively stable for women across the age groups (18.8%, 16.3%, 17.5% and 19.3%), while for men the prevalence peaked in the 75 to 79 age group and then declined (29.1%, 32.1%, 40.1%, 34.4%). Based on these figures, low numbers of older people are taking the opportunity to engage in this safe and accessible form of physical activity. Targeting older people at age 65 may set the stage for maintaining higher rates throughout old age.
New estimates of disease prevalence in older Australians
The large DYNOPTA dataset and availability of sample weights has allowed the estimation of prevalence of chronic diseases for which Australia currently lacks reliable and sensitive information in the older adult age range. The exclusive focus of the contributing DYNOPTA studies on older adulthood provides sufficient numbers of individuals in the older age ranges to estimate disease prevalence in narrow 5-year age bands. We have conducted two studies thus far evaluating the relative strengths of the DYNOPTA sample by comparing our prevalence estimates with those reported by surveys conducted by the Australian Bureau of Statistics. The first study (14) compared the estimates of probable dementia and possible cognitive impairment using DYNOPTA with those reported by the 1997 and 2007 National Surveys of Mental Health and Wellbeing (NSMHW)(15,16). DYNOPTA estimates were, lower than the values in the NSMHWs, and were more similar to those reported in other meta-analyses based on Australian and European data. Figure 1 shows the rates of probable dementia in DYNOPTA. Because such information is used to project future disease prevalence and thus allocation of health-care spending, accurate projections are vital, especially given that small differences in prevalence rates can have large implications when projected up to the population.
Figure 1.
Percentage of sample with probably dementia (Mini-Mental State Examination <24) by age and gender in DYNOPTA (weighted)
We similarly compared the prevalence of chronic medical conditions among Australian older adults (17) with those reported by the 2001 National Health Survey(18). Our observed prevalence rates were lower than those reported by NHS for diabetes, but higher for asthma and hypertension. Further, because the NHS only provided data across three older adult age bands, information was missed that showed the prevalence of disease did not always increase with older age. For example, according to the NHS, arthritis rates appear to increase with age with the upper age category being 75+, but in DYNOPTA arthritis is less prevalent in the mid-70s and beyond compared to the 60-64, 65-69, 70-74 age categories.
Novel developments in understanding trajectories of physical disability
Mobility limitations are a common form of disability in late life with musculoskeletal conditions being related to a wide range of risk factors. Little is known about the changes in mobility limitations over time as people age and the factors that alter this trajectory. Analyses of DYNOPTA data have shown that at baseline, younger age groups have close to 100% probability of being able to walk at least 1km. This probability declines steadily with increasing baseline age up until the late 80s. It seems that individuals recruited to these studies at older ages tended to be exceptionally fit, particularly among the men. However, participants who commenced the studies at an advanced age were also less likely to continue in the studies. Over time, there is little change in walking for younger age ranges, although their ability to walk 1km does decline slightly in the most recent surveys. Older adults, however, show greater variability and a much steeper decline in walkability. The ability to walk 1km declines more steeply with both age and ageing for women than for men(19).
Characteristics of Driving participation
Driving is a key component to the health and well-being of older adults. The DYNOPTA project provides a unique opportunity to investigate the cognitive, sensory, physical and psychological health of older Australian drivers, as well as non-drivers. We investigated the demographic, cognitive, sensory and physical health (via medical conditions) between older Australian drivers and non-drivers (n = 5206) using DYNOPTA (20). Men and participants with higher-level occupations were more likely to be drivers, while marital status, older age, more medical conditions, and poorer vision increased the odds of not driving. Descriptive analyses revealed a large proportion of men who reported driving with probable visual or cognitive impairments. A second investigation of driving in DYNOPTA examined the impact of age-based Australian medical and visual-based licensure renewal policies on driving status(21). This work showed that Australians living in states with age-based testing were less likely to continue driving after the age corresponding to the age-based testing Future investigations will take advantage of the longitudinal data within DYNOPTA to investigate such driving-related issues over time and between cohorts of older adults.
Social relationships in the oldest-old
Positive social relationships are now recognised as fundamental for physical and mental health across the lifespan(22). DYNOPTA provides several strengths that allow for a more comprehensive examination of the antecedents and consequences of social network characteristics in late life than has previously been undertaken in the Australian context. These include examination of social engagement among the oldest old, and social network characteristics of low prevalence groups in the population who may be at increased risk for social isolation, such as never-married older adults in poor health. Future DYNOPTA investigations will examine the role of social network characteristics in predicting key outcomes including depressive symptoms and mortality.
Evaluation of prevalence of mental disorders in late-life
Mental health and well-being is one of the key areas of the DYNOPTA project and all of the contributing DYNOPTA studies had used some measure of depression or mental health, including the Short-Form Health Survey-36 (SF-36), the Short-Form Health Survey-12 (SF-12), the Centre for Epidemiological Studies Depression Scale (CESD), and the Psychogeriatric Assessment Scales (PAS). Development of a common metric for depression measures across the participating studies has been developed through use of standardized scores validated against population prevalence rates(24). Again, the strength of the DYNOPTA project lies in its capacity to compile data for older Australians who are typically under-represented in national surveys, including the NSMHWB. Table 4 shows the DYNOPTA sample is considerably more sizable than the sample included in the NSMHWB surveys which typically suggest a decline in mental health with increasing age. Analysis of the weighted DYNOPTA data allows for the more complex non-linear patterns of mental health among older participants (75+) to be modelled, which is not possible in the NSMHWB.
Table 4.
Composition of DYNOPTA sample and National Mental Health Surveys according to age-group and gender
| Age Groups | DYNOTPA | NMHS 1997 | NHMS 2007 | |||
|---|---|---|---|---|---|---|
|
| ||||||
| Males | Females | Males | Females | Males | Females | |
| 45-49 | 1,271 | 14763 | 456 | 528 | 295 | 337 |
| 50-54 | 1238 | 1758 | 359 | 489 | 271 | 361 |
| 55-59 | 1089 | 1246 | 351 | 336 | 288 | 375 |
| 60-64 | 2171 | 2168 | 276 | 316 | 316 | 294 |
| 65-69 | 1218 | 1493 | 285 | 293 | 333 | 303 |
| 70-74 | 1488 | 13406 | 201 | 299 | 241 | 227 |
| 75+ | 3092 | 4251 | 245 | 469 | 330 | 471 |
| 80+ | 1853 | 2266 | - | - | 132 | 225 |
CONTRIBUTION TO SOCIAL AND ECONOMIC PLANNING FOR THE FUTURE
DYNOPTA also supports the development of a microsimulation model that will estimate the prevalence and cost of the future disease burden of the baby-boomer cohort (DYNOPTASIM). This unique model will be based on the transition rates obtained from longitudinal data available in the DYNOPTA dataset, in contrast to many similar models that use only cross-sectional information. The DYNOPTA project has already contributed evidence used in dementia projections contributed to the evaluation of licensing policies in Australia, and a number the projects in process provide information on prevalence that can inform health planning.
FUTURE DIRECTIONS FOR DYNOPTA
DYNOPTA has only been possible through strategic funding for a project that lies outside the traditional prototype of a project grant, genuine multidisciplinary and multicentre collaboration, and the development of new methods. Thus far, we have produced a number of substantive outputs in relation to methodological advancements and health-related research. In the future, we anticipate a substantial dividend in terms of policy relevant publications and providing a strong evidence base for both individual and population ageing in Australia.
Key Points.
DYNOPTA is a study that has pooled datasets from nine Australian Longitudinal Studies of Ageing
DYNOPTA provides new estimates of the prevalence of disease and disability based on larger numbers of the oldest-old than has previously been possible
DYNOPTA allows for the evaluation of current health and social policy in areas such as alcohol consumption, and licensing
DYNOPTA is being used to generate projections of expected health and morbidity within the baby-boomer cohort.
Acknowledgments
The DYNOPTA Team include H. Booth, A. Broe, L. Brown, P. Butterworth, R. Cumming, R. Gibson, J. Healy, M.A. Luszcz, P. Mitchell and J. Shaw. The data on which this research is based were drawn from several Australian longitudinal studies including: the Australian Longitudinal Study of Ageing (ALSA), the Australian Longitudinal Study on Women’s Health (ALSWH), the Australian Diabetes, Obesity and Lifestyle Study (AusDiab), the Blue Mountains Eye Study (BMES), the Canberra Longitudinal Study of Ageing (CLS), the Household, Income and Labour Dynamics in Australia study (HILDA), the Melbourne Longitudinal Studies on Healthy Ageing (MELSHA), the Personality AndTotal Health Through Life Study (PATH) and the Sydney Older Persons Study (SOPS). These studies were pooled and harmonised for the DYNOPTA project. All studies would like to thank the participants for volunteering their time to be involved in the respective studies. Details of all studies contributing data to DYNOPTA, including individual study leaders and funding sources, are available on the DYNOPTA website (http://DYNOPTA.anu.edu.au). The findings and views reported in this paper are those of the author(s) and not those of the original studies or their respective funding agencies. This work was supported by the National Health and Medical Research Council (NHMRC) Grant # 410215. Professor Anstey is funded by NHMRC Fellowship #366756 and has received a speaker fee from Pfizer in the past two years. Allison Bielak was supported by a postdoctoral research fellowship from the Canadian Institutes of Health Research. Dr Ross is supported by the UAB Edward R. Roybal Center for Translational Research on Aging and Mobility, NIA 2 P30 AG022838.
References
- 1.Booth H, Tickle L. The future aged: new projections of Australia’s elderly population. Australasian Journal on Ageing. 2003;22:38–44. [Google Scholar]
- 2.NRC. Preparing for an ageing world: the case for cross-national research. National Research Council; 2001. World PoaRAaNDfaA, ed. [Google Scholar]
- 3.Andrews K. National Strategy for an Ageing Australia An Older Australia, Challenges and Opportunities for all. Canberra: The Australian Government Publisher; 2003. [Google Scholar]
- 4.DHAC. An active way to better health: National physical activity guidelines for adults. Care DoHaA, ed: Australian Government; 1999. [Google Scholar]
- 5.NHMRC. Australian Alcohol Guidelines: Health risks and benefits. Canberra: National Health and Medical Research Council; 2001. [Google Scholar]
- 6.Baltes PB, Staudinger UM, Lindenberger U. Lifespan psychology: theory and application to intellectual functioning. Annu Rev Psychol. 1999;50:471–507. doi: 10.1146/annurev.psych.50.1.471. [DOI] [PubMed] [Google Scholar]
- 7.Kuh D, Ben-Shlomo Y, Lynch J, et al. Life course epidemiology. J Epidemiol Community Health. 2003;57:778–83. doi: 10.1136/jech.57.10.778. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Anstey KJ, Byles JE, Luszcz MA, et al. Cohort profile: The Dynamic Analyses to Optimize Ageing (DYNOPTA) project. Int J Epidemiol. 2009;39:44–51. doi: 10.1093/ije/dyn276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Byles J, Dobson A. Increasing returns: the value of time in longitudinal studies on ageing. Lessons from the Australian Longitudinal Study on Women’s Health. Australasian Journal on Ageing. doi: 10.1111/j.1741-6612.2011.00531.x. in submission. [DOI] [PubMed] [Google Scholar]
- 10.Kiely KM, Mitchell P, Anstey KJ. The case for conditional single imputation as preferable to other imputation methods in estimating study censored demographic data in in a pooled national dataset. Journal of Clinical Epidemiology. in submission. [Google Scholar]
- 11.Burns RA, Butterworth P, Kiely K, et al. Multiple imputation is an efficient method for harmonising the Mini-Mental State Examination with missing item-level data. Journal of Clinical Epidemiology. doi: 10.1016/j.jclinepi.2010.10.011. in press. [DOI] [PubMed] [Google Scholar]
- 12.Graham JW. Missing data analysis: making it work in the real world. Annu Rev Psychol. 2009;60:549–76. doi: 10.1146/annurev.psych.58.110405.085530. [DOI] [PubMed] [Google Scholar]
- 13.Burns RA, Birrell CL, Steel D, et al. Alcohol and smoking behaviours in older Australian Adults in the DYNOPTA Project: Prevalence, cohort effects and sociodemographic differentials. Int J Epidemiol. doi: 10.1007/s00127-012-0558-x. [DOI] [PubMed] [Google Scholar]
- 14.Anstey KJ, Burns RA, Birrell CL, et al. Estimates of probable dementia prevalence from population-based surveys compared with dementia prevalence estimates based on meta-analyses. BMC Neurol. 2010;10:62. doi: 10.1186/1471-2377-10-62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.McLennan W. ABS. National Survey of Mental Health and Wellbeing of Adults: Users’ Guide. Canberra: Australian Bureau of Statistics; 1999. [Google Scholar]
- 16.Pink B. National Survey of Mental Health and Wellbeing: Users Guide. Canberra: Australian Bureau of Statistics; 2007. [Google Scholar]
- 17.Bielak AAM, Byles JE, Luszcz MA, Anstey KJ. Prevalence of Disease Among Over 12, 000 Australian Older Adults Between 2000-2002: Results from the DYNOPTA Project. Journal of Clinical Epidemiology. in submission. [Google Scholar]
- 18.Australian, Bureau, of, Statistics. National health survey: Summary of results, 2001. Canberra: 2002. [Google Scholar]
- 19.Byles J, Gibson R, Humphreys M. Age associated decline in physical mobility: Trends and transitions. Australasian Journal on Ageing. 2009;28:47. [Google Scholar]
- 20.Ross LA, Anstey KJ, Kiely KM, et al. Older drivers in Australia: trends in driving status and cognitive and visual impairment. J Am Geriatr Soc. 2009;57:1868–73. doi: 10.1111/j.1532-5415.2009.02439.x. [DOI] [PubMed] [Google Scholar]
- 21.Ross L, Browning C, Mitchell P, et al. Age-Based Testing for Driver’s License Renewal: Potential Implications for Older Australians. Journal of the American Geriatrics Society. doi: 10.1111/j.1532-5415.2010.03241.x. in press. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Holt-Lunstad J, Smith TB, Layton JB. Social relationships and mortality risk: a meta-analytic review. PLoS Med. 7:e1000316. doi: 10.1371/journal.pmed.1000316. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Windsor TD, Bielak AAM, Anstey KJ, et al. Resource availability and social engagement in older adulthood: Results from DYNOPTA. Ageing and Society. in submission. [Google Scholar]
- 24.Burns RA, Butterworth P, Luszcz MA, et al. Developing a common depression variable for the DYNOPTA project to describe gender and age effects on rates of probable depression in large samples of older Australians. Australian and New Zealand Journal Psychiatry. in submission. [Google Scholar]