Estimating dementia prevalence using remote diagnoses and algorithmic modelling: a population-based study of a rural region in South Africa

Meagan T Farrell; Darina T Bassil; Muqi Guo; M Maria Glymour; Ryan G Wagner; Stephen Tollman; Kenneth M Langa; Adam M Brickman; Jennifer J Manly; Lisa F Berkman

doi:10.1016/S2214-109X(24)00325-5

Estimating dementia prevalence using remote diagnoses and algorithmic modelling: a population-based study of a rural region in South Africa

Lancet Glob Health. 2024 Dec;12(12):2003-2011. doi: 10.1016/S2214-109X(24)00325-5.

Authors

Affiliations

¹ Harvard Center for Population and Development Studies, Harvard University, Cambridge, MA, USA. Electronic address: mfarrell@hsph.harvard.edu.
² Harvard Center for Population and Development Studies, Harvard University, Cambridge, MA, USA.
³ Harvard Center for Population and Development Studies, Harvard University, Cambridge, MA, USA; Department of Global Health and Population, Harvard T H Chan School of Public Health, Harvard University, Cambridge, MA, USA.
⁴ Department of Epidemiology, Boston University School of Public Health, Boston, MA, USA.
⁵ MRC/Wits Rural Public Health and Health Transitions Research Unit (Agincourt), School of Public Health, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa.
⁶ MRC/Wits Rural Public Health and Health Transitions Research Unit (Agincourt), School of Public Health, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa; INDEPTH Network, Accra, Ghana.
⁷ Institute for Healthcare Policy and Innovation and Institute for Social Research, University of Michigan, Ann Arbor, MI, USA; Department of Internal Medicine, University of Michigan Medical School, Ann Arbor, MI, USA; Veterans Affairs Center for Clinical Management Research, Ann Arbor, MI, USA.
⁸ Taub Institute for Research on Alzheimer's Disease and the Aging Brain, Gertrude H Sergievsky Center, and Department of Neurology, Vagelos College of Physicians and Surgeons, Columbia University, New York, NY, USA.
⁹ Harvard Center for Population and Development Studies, Harvard University, Cambridge, MA, USA; Department of Social and Behavioral Sciences, Harvard T H Chan School of Public Health, Harvard University, Cambridge, MA, USA; MRC/Wits Rural Public Health and Health Transitions Research Unit (Agincourt), School of Public Health, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa.

PMID: 39577957
DOI: 10.1016/S2214-109X(24)00325-5

Abstract

Background: Dementia is a leading cause of global death and disability. High-quality data describing dementia prevalence and burden remain scarce in sub-Saharan Africa. Health and Aging in Africa: A Longitudinal Study in South Africa (HAALSI) fills evidence gaps with longitudinal data on cognition, biomarkers, and everyday function in a population-based cohort of Black South Africans, aged 40 years and older, in a rural subdistrict. This study uses consensus diagnoses and prediction algorithms to estimate dementia prevalence.

Methods: Data were from eligible HAALSI Wave 2 respondents aged 50 years or older (n=3662) and were collected between September, 2019, and January, 2020. An enriched sub-cohort (ie, including a high proportion of individuals with cognitive impairment; n=632) completed a battery of rigorous neuropsychological and clinical assessments and received expert classification of cognitively unimpaired, mild cognitive impairment, or dementia. Logistic regression was used to predict dementia status within the sub-cohort using predictor variables from the parent HAALSI wave. Coefficients were applied to the parent cohort to obtain dementia probability scores and calculate dementia prevalence. Optimal probability cut points to classify individual cases were selected for each model.

Findings: When the sub-cohort was reweighted to reflect the full HAALSI population, the estimated prevalence of dementia was 18% (95% CI 15-22), with steep age gradients. Four models of increasing complexity showed good discrimination between dementia and non-dementia (area under receiver operating characteristic curves 0·78-0·84; classification accuracy 74-81%). Model-based dementia prevalence estimates aligned closely with weighted prevalence; model performance was consistent in cross-validated datasets.

Interpretation: HAALSI is among the first studies to use algorithmic methods to describe dementia prevalence in a population-based sample in South Africa. These efforts could provide a foundation to expand understanding of dementia epidemiology in a region of the world experiencing rapid population ageing.

Funding: National Institute on Aging.

MeSH terms

Aged
Aged, 80 and over
Algorithms*
Dementia* / diagnosis
Dementia* / epidemiology
Female
Humans
Longitudinal Studies
Male
Middle Aged
Prevalence
Rural Population* / statistics & numerical data
South Africa / epidemiology