Abstract dementia risk prediction model development in low‐ and middle‐income countries: The 10/66 study: Epidemiology / Risk and protective factors in MCI and dementia. (7th December 2020)

• Record Type: Journal Article
• Authors: Worrall, Alice Louise; Prina, Matthew; Pakpahan, Eduwin; Siervo, Mario; Terrera, Graciela Muniz; Mohan, Devi; Acosta, Daisy; Pichardo, Guillermina Rodriguez; Sosa‐Ortiz, Ana Luisa; Acosta‐Castillo, Gilberto Isaac; Llibre, Juan; Prince, Martin J; Robinson, Louise; Stephan, Blossom CM
• Is Part Of: Alzheimer's & dementia. Volume 16(2020)Supplement 10
• Page Start: n/a
• Page End: n/a
• Publication Date: 2020-12-07
• Subjects: Alzheimer's disease -- Periodicals; Alzheimer Disease -- Periodicals; Dementia -- Periodicals; Démence; Maladie d'Alzheimer; Périodique électronique (Descripteur de forme); Ressource Internet (Descripteur de forme); 616.83
• DOI: 10.1002/alz.042071
• Languages: English
• ISSNs: 1552-5260

Abstract

Abstract: Background: Most people with dementia live in Low and Middle Income Countries (LMICs), where little research on dementia risk prediction modelling has occurred. This study aimed to develop new models to simply predict all‐cause dementia, suitable for use in LMICs. Country‐specific models were expected, due to different risk profiles. Method: Data was from the 10/66 cohort study. Individuals aged ≥65 years without dementia at baseline (N=11, 143) were recruited from Cuba, the Dominican Republic, Peru, Venezuela, Mexico, Puerto Rico and China. Dementia incidence was assessed over a mean follow‐up of 3.8 years (SD=1.3 years). Variables were selected to be tested that have been associated with dementia previously. Two Cox risk models were produced for each site, with and without objective cognitive variables. Predictive accuracy (c‐statistic) and calibration were tested. Result: 1, 069 individuals progressed to dementia during follow‐up. The models had mostly moderate to good predictive accuracy. The models in the total sample performed moderately (cognitive model c‐statistic = 0.74; 95%CI: 0.72‐0.75; non‐cognitive model c‐statistic = 0.71; 95%CI: 0.70‐0.73). In each country, the cognitive models' c‐statistics ranged from 0.70 (95%CI: 0.67‐0.74) in China to 0.84 (95%CI: 0.80‐0.88) in Peru, and the non‐cognitive models' c‐statistics ranged from 0.67 (95%CI: 0.63‐0.71) in the Dominican Republic to 0.80 (95%CI: 0.74‐0.85) in Peru. There were no major noticeable patterns of variables included in each country‐specific model. Model calibration was however poor. Conclusion: Different prediction models were necessary for each country, most of which had moderate to good predictive accuracy. Further research is needed to improve calibration, and to determine whether risk prediction is cost‐effective, ethical and acceptable in LMICs. Dementia risk prediction is important so that individuals at high risk can be identified, and their risk factors addressed, to help decrease the burden of dementia.