Limits...
Reliability of Nationwide Prevalence Estimates of Dementia: A Critical Appraisal Based on Brazilian Surveys.

Chaimowicz F, Burdorf A - PLoS ONE (2015)

Bottom Line: Among the 35 studies found, 15 analyzed population-based random samples.However, after adjustment for accuracy of screening, the best available evidence points towards a figure between 15.2% and 16.3%.Similar evaluations regarding other developing countries are needed in order to verify the scope of these limitations.

View Article: PubMed Central - PubMed

Affiliation: Department of Internal Medicine, Faculty of Medicine, Federal University of Minas Gerais, Belo Horizonte, Minas Gerais, Brazil; Department of Public Health, Erasmus Medical Center, Erasmus University, Rotterdam, Netherlands.

ABSTRACT

Background: The nationwide dementia prevalence is usually calculated by applying the results of local surveys to countries' populations. To evaluate the reliability of such estimations in developing countries, we chose Brazil as an example. We carried out a systematic review of dementia surveys, ascertained their risk of bias, and present the best estimate of occurrence of dementia in Brazil.

Methods and findings: We carried out an electronic search of PubMed, Latin-American databases, and a Brazilian thesis database for surveys focusing on dementia prevalence in Brazil. The systematic review was registered at PROSPERO (CRD42014008815). Among the 35 studies found, 15 analyzed population-based random samples. However, most of them utilized inadequate criteria for diagnostics. Six studies without these limitations were further analyzed to assess the risk of selection, attrition, outcome and population bias as well as several statistical issues. All the studies presented moderate or high risk of bias in at least two domains due to the following features: high non-response, inaccurate cut-offs, and doubtful accuracy of the examiners. Two studies had limited external validity due to high rates of illiteracy or low income. The three studies with adequate generalizability and the lowest risk of bias presented a prevalence of dementia between 7.1% and 8.3% among subjects aged 65 years and older. However, after adjustment for accuracy of screening, the best available evidence points towards a figure between 15.2% and 16.3%.

Conclusions: The risk of bias may strongly limit the generalizability of dementia prevalence estimates in developing countries. Extrapolations that have already been made for Brazil and Latin America were based on a prevalence that should have been adjusted for screening accuracy or not used at all due to severe bias. Similar evaluations regarding other developing countries are needed in order to verify the scope of these limitations.

No MeSH data available.


Related in: MedlinePlus

Age-specific prevalence of dementia among the studies selected for full analysis.*.*Magalhães: data not available. Herrera adjusted: prevalence in the group aged 85 years and older (88.7%) is not presented.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4488471&req=5

pone.0131979.g003: Age-specific prevalence of dementia among the studies selected for full analysis.*.*Magalhães: data not available. Herrera adjusted: prevalence in the group aged 85 years and older (88.7%) is not presented.

Mentions: Fig 3 shows that among the six studies, the higher the age group, the higher the prevalence of dementia (S2 Table additionally presents non-adjusted prevalence from the studies of Bottino, Lopes and Herrera). However, differences among surveys regarding age-specific prevalence are relevant. The prevalence at 70–74 years of age found by Ramos-Cerqueira and Bottino were, respectively, 0.1% and 7.1%, whereas the adjusted prevalence at 80–84 years of age found by Bottino and Lopes—who used a similar methodology—were 16.1% and 24.0%. The differences within each survey regarding age variation were also considerable. While the prevalence at 65–69 years and 70–74 years did not vary in the surveys from Scazufca (2.3% and 2.0%) and Ramos-Cerqueira (0.1%), it doubled in the survey from Herrera (1.6% to 3.2%) and almost tripled in the survey from Lopes (2.2% to 6.3%). The prevalence found by Scazufca among those aged 85years and older (21.4%) was 10 times that found among those aged 70–74 years (2.0%), but was only three times higher in the study from Bottino (22.4% and 7.1%).


Reliability of Nationwide Prevalence Estimates of Dementia: A Critical Appraisal Based on Brazilian Surveys.

Chaimowicz F, Burdorf A - PLoS ONE (2015)

Age-specific prevalence of dementia among the studies selected for full analysis.*.*Magalhães: data not available. Herrera adjusted: prevalence in the group aged 85 years and older (88.7%) is not presented.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4488471&req=5

pone.0131979.g003: Age-specific prevalence of dementia among the studies selected for full analysis.*.*Magalhães: data not available. Herrera adjusted: prevalence in the group aged 85 years and older (88.7%) is not presented.
Mentions: Fig 3 shows that among the six studies, the higher the age group, the higher the prevalence of dementia (S2 Table additionally presents non-adjusted prevalence from the studies of Bottino, Lopes and Herrera). However, differences among surveys regarding age-specific prevalence are relevant. The prevalence at 70–74 years of age found by Ramos-Cerqueira and Bottino were, respectively, 0.1% and 7.1%, whereas the adjusted prevalence at 80–84 years of age found by Bottino and Lopes—who used a similar methodology—were 16.1% and 24.0%. The differences within each survey regarding age variation were also considerable. While the prevalence at 65–69 years and 70–74 years did not vary in the surveys from Scazufca (2.3% and 2.0%) and Ramos-Cerqueira (0.1%), it doubled in the survey from Herrera (1.6% to 3.2%) and almost tripled in the survey from Lopes (2.2% to 6.3%). The prevalence found by Scazufca among those aged 85years and older (21.4%) was 10 times that found among those aged 70–74 years (2.0%), but was only three times higher in the study from Bottino (22.4% and 7.1%).

Bottom Line: Among the 35 studies found, 15 analyzed population-based random samples.However, after adjustment for accuracy of screening, the best available evidence points towards a figure between 15.2% and 16.3%.Similar evaluations regarding other developing countries are needed in order to verify the scope of these limitations.

View Article: PubMed Central - PubMed

Affiliation: Department of Internal Medicine, Faculty of Medicine, Federal University of Minas Gerais, Belo Horizonte, Minas Gerais, Brazil; Department of Public Health, Erasmus Medical Center, Erasmus University, Rotterdam, Netherlands.

ABSTRACT

Background: The nationwide dementia prevalence is usually calculated by applying the results of local surveys to countries' populations. To evaluate the reliability of such estimations in developing countries, we chose Brazil as an example. We carried out a systematic review of dementia surveys, ascertained their risk of bias, and present the best estimate of occurrence of dementia in Brazil.

Methods and findings: We carried out an electronic search of PubMed, Latin-American databases, and a Brazilian thesis database for surveys focusing on dementia prevalence in Brazil. The systematic review was registered at PROSPERO (CRD42014008815). Among the 35 studies found, 15 analyzed population-based random samples. However, most of them utilized inadequate criteria for diagnostics. Six studies without these limitations were further analyzed to assess the risk of selection, attrition, outcome and population bias as well as several statistical issues. All the studies presented moderate or high risk of bias in at least two domains due to the following features: high non-response, inaccurate cut-offs, and doubtful accuracy of the examiners. Two studies had limited external validity due to high rates of illiteracy or low income. The three studies with adequate generalizability and the lowest risk of bias presented a prevalence of dementia between 7.1% and 8.3% among subjects aged 65 years and older. However, after adjustment for accuracy of screening, the best available evidence points towards a figure between 15.2% and 16.3%.

Conclusions: The risk of bias may strongly limit the generalizability of dementia prevalence estimates in developing countries. Extrapolations that have already been made for Brazil and Latin America were based on a prevalence that should have been adjusted for screening accuracy or not used at all due to severe bias. Similar evaluations regarding other developing countries are needed in order to verify the scope of these limitations.

No MeSH data available.


Related in: MedlinePlus