RESUMO
This paper uses five strategies to evaluate the reliability and other measurement qualities of the Ten Questions screen for childhood disability. The screen was administered for 22,125 children, aged 2-9 years, in Bangladesh, Jamaica and Pakistan. The test-retest approach involving small sub-samples was useful for assessing reliability of overall screening results, but not of individual items with low prevalence. Alternative strategies focus on the internal consistency and structure of the screen as well as item analyses. They provide evidence of similar and comparable qualities of measurement in the three culturally divergent populations, indicating that the screen is likely to produce comparable data across cultures. One of the questions, however, correlates with the other questions differently in Jamaica, where it appears to "over-identify" children as seriously disabled. The methods and findings reported here have general applications for the design and evaluation of questionnaires for epidemiologic research, particularly when the goal is to gather comparable data in geographically and culturally diverse settings.
Assuntos
Comparação Transcultural , Deficiências do Desenvolvimento/diagnóstico , Avaliação da Deficiência , Bangladesh , Criança , Pré-Escolar , Deficiências do Desenvolvimento/etnologia , Métodos Epidemiológicos , Análise Fatorial , Feminino , Humanos , Jamaica , Masculino , Paquistão , Reprodutibilidade dos Testes , Inquéritos e QuestionáriosRESUMO
An international study to validate the Ten Questions screen for serious childhood disability was undertaken in communities in Bangladesh, Jamaica, and Pakistan, where community workers screened more than 22,000 children ages 2-9 years. All children who screened positive, as well as random samples of those who screened negative, were referred for clinical evaluations. Applying comparable diagnostic criteria, the sensitivity of the screen for serious cognitive, motor, and seizure disabilities is acceptable (80-100%) in all three populations, whereas the positive predictive values range from 3 to 15%. These results confirm the usefulness of the Ten Questions as a low-cost and rapid screen for these disabilities, although not for vision and hearing disabilities, in populations where few affected children have previously been identified and treated. They also show that the value of the Ten Questions for identifying disability in underserved populations is limited to that of a screen; more thorough evaluations of children screened positive are necessary to distinguish true- from false-positive results and to identify the nature of the disability if present.