Estimating the Incidence of Dementia: The Impact of Adjusting for Subject Attrition Using Health Care Utilization Data
Introduction
Incidence studies of dementia often rely entirely on follow-up assessments to identify cases, and subjects who do not complete a follow-up assessment are often omitted from the analysis. These subjects, however, cannot be assumed to be missing at random 1, 2 because dementia decreases life expectancy (3), and subjects who develop dementia may thus be more likely to have been too sick to participate or to have died before the follow-up examination. Selective attrition due to excess mortality and morbidity has long been recognized (4). A recent systematic review of longitudinal studies in older adults showed that older age and cognitive impairment—key factors in studies of dementia—were the primary and most consistent predictors of attrition (5). Although longitudinal studies of dementia routinely employ strategies to minimize attrition, such as shorter intervals between assessments or data collection from alternative sources, loss of subjects due to death is inevitable and can be substantial (e.g., from 7.0% 6, 7 to 22.4% (8)). Subjects are also lost to follow up because they refuse to participate after their baseline assessment, cannot be located, have moved out of the study area, or are otherwise unable to complete the study. It is unclear how the incidence of dementia among these subjects would compare to subjects who participate in a follow-up assessment, although this type of attrition has been shown to result in underestimates of cognitive decline (9).
Excluding subjects lost to follow up may affect the accuracy of dementia incidence rates and, in turn, hinder the utility of these rates to provide valid projections of future service needs. To assess the impact of attrition, information on dementia status is required for all subjects regardless of follow-up status, and a source of information common to all study subjects, such as health care utilization records, is preferable.
The objective of this study was to estimate the incidence of dementia and the influence of subject attrition. Assessment of incidence rates based on follow-up examinations and those based on health care utilization records (which are not affected by subject attrition) allowed estimation of the impact of loss to follow up and adjustment of incidence rates for attrition.
Section snippets
Study Population
The Manitoba Study of Health and Aging (MSHA) is a population-based longitudinal study of aging and dementia in community-based and institutionalized adults 65 years of age or older. The MSHA study design (Fig. 1) has been described elsewhere 10, 11. Briefly, in 1991, the population of Manitoba aged 65 years and older was 147,372, of whom 139,579 lived in the community and 7793 in institutions (nursing homes). Baseline data were collected on 1763 community and 189 institutional subjects in
Incidence of Dementia Based on Follow-up Assessments
Approximately 5 years after baseline, 117 subjects were diagnosed with dementia at follow-up clinical assessment among community subjects. Age-specific incidence rose from 4.3/1000 person-years to 79.6/1000 person-years from the age groups 65–69 years to 90 years and older, respectively (Table 1). The age- and gender-standardized incidence of dementia was 15.6/1000 person-years (95% confidence interval [CI: 12.4–18.7]). There was no significant difference in incidence of dementia between men
Discussion
Incidence rates for dementia and AD were calculated for a population-based sample 65 years or older from both the community and institutions. We considered follow-up examination and clinical assessment as the gold standard method to identify cases of dementia. Incidence estimates from this method, however, are potentially biased because completers may not be representative of the total population at risk. A unique feature of our study was the availability of health care utilization records,
Conclusions
We have shown that censoring of decedents biases incidence rates, resulting in underestimates of dementia incidence in our study. Loss to follow up of subjects who refused to participate or who were unavailable at follow up also produced underestimates of dementia incidence, but attrition of these subjects had less impact than did attrition of decedents. Sources of data available for all subjects, such as health care utilization records, can be used to correct incidence estimates for attrition
References (36)
- et al.
A systematic literature review of attrition between waves in longitudinal studies in the elderly shows a consistent pattern of dropout between differing studies
J Clin Epidemiol
(2005) - et al.
Comparison of survey and physician claims data for detecting hypertension
J Clin Epidemiol
(1997) Rates of senile dementia, Alzheimer's type, in the Baltimore Longitudinal Study
J Chron Dis
(1986)A shared random effect parameter approach for longitudinal dementia data with non-ignorable missing data
Statist Med
(2004)- et al.
Population-based studies of ad: Message and methods: An epidemiologic view
Statist Med
(2004) - et al.
Clinical progression of dementia study group: A reevaluation of the duration of survival after the onset of dementia
N Engl J Med
(2001) Methodological concerns: Longitudinal studies of dementia
Alzheimer Dis Assoc Disord
(1987)- et al.
Incidence and risk of dementia: The Rotterdam Study
Am J Epidemiol
(1998) - et al.
Incidence of dementia and Alzheimer's disease in elderly community residents of South-Western France
Int J Epidemiol
(1994) The incidence of dementia in Canada
Neurology
(2000)
Estimating the true extent of cognitive decline in the old old
J Am Geriatr Soc
Risk factors for Alzheimer's disease: A population-based, longitudinal study in Manitoba, Canada
Int J Epidemiol
Canadian study of health and aging: Study methods and prevalence of dementia
Can Med Assoc J
The Modified Mini-Mental State (3MS) Examination
J Clin Psychiatry
Effects of screening errors and differential mortality on the estimation of the incidence of dementia in the Canadian Study of Health and Aging
Int Psychogeriatr
Diagnostic and Statistical Manual of Mental Disorders, Revised (DSM III-R)
Clinical diagnosis of Alzheimer's disease: Report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer's Disease
Neurology
Cited by (33)
Impact of age, comorbidity, and polypharmacy on receipt of systemic therapy in advanced cancers: A retrospective population-based study
2024, Journal of Geriatric OncologyThe impact of methods to handle missing data on the estimated prevalence of dementia and mild cognitive impairment in a cross-sectional study including non-responders
2017, Archives of Gerontology and GeriatricsCitation Excerpt :This conclusion is consistent with a Brazilian study that reported prevalence was increased by 111.86%, from 5.9% to 12.5%, after correction for non-response bias (Lopes, Ferrioli, Nakano, Litvoc, & Bottino, 2012). A similar trend was also found regarding the incidence of dementia, with supporting evidence from a Canadian study (Tyas et al., 2006) and from Zhang MY’s report based on mainland China (Zhang et al., 1998Zhang, Chen, Katzman, Yu, & Liu, 1998). A prediction correction model incorporating proxy assessments of the cognitive function of non-responders was also previously reported to increase the estimated prevalence of dementia by 12% (Wu et al., 2013).
Risk Factors for Mild Cognitive Impairment, Dementia and Mortality: The Sydney Memory and Ageing Study
2017, Journal of the American Medical Directors AssociationCharacteristics of participants' and caregivers' influence on non-response in a cross-sectional study of dementia in an older population
2016, Archives of Gerontology and GeriatricsCitation Excerpt :It is reliable and convenient to collect such participants’ detailed, long-term medical records and describe the characteristics of their caregivers (Tan et al., 2014). Therefore, the majority of the baseline information could easily be obtained from the veterans’ healthcare system; the non-response rate in phase 1 was only 5.4%, which was lower than the majority of the dementia studies in older adults or the oldest old (Boersma et al., 1997; Helliwell et al., 2001; Lucca et al., 2011; Miyamotoa et al., 2009; Chatfield et al., 2005; Tyas et al., 2006; Lopes et al., 2012; Bootsma-van der Wiel et al., 2002). Support and assistance from the veterans’ communities persisted, despite the increase (to 35.0%) in the non-response rate in phase 2.
Mortality with Alzheimer's disease and dementia in France, 2006
2010, Revue d'Epidemiologie et de Sante Publique
The Manitoba Study of Health and Aging (MSHA) was funded primarily by Manitoba Health (1990–1993) and Manitoba Health's Healthy Communities Development Fund (1995–1999). Additional funding was provided through the Canadian Study of Health and Aging (CSHA) by the Seniors Independence Research Program of the National Health Research and Development Program of Health Canada (Project No. 6606-3954-MC[S]). The results and conclusions are those of the authors and no official endorsement by Manitoba Health is intended or should be inferred.