Abstract Changes in demographic patterns, as well as improvements in health systems, increase the elderly population in the world and especially in Latin America. These processes increase the prevalence of chronic and degenerative diseases such as depression. Depression involves social and financial costs that can cause disorders to the health and daily life of those who suffer from it and those around them, such as high morbidity and mortality. This study uses classical test theory and Rasch model to seek for psychometric evidence to support the suitability of Yesavage's short-scale questionnaire (GDS-15) as a suitable instrument to identify depressive symptoms in general older adult populations in Latin America. The unidimensionality of the scale in each country was validated by performing a factorial analysis of principal components with a tetrachoric correlation matrix. Cronbach's Alpha was also calculated as an indicator of the internal validity of the scale and in order to provide the values of this indicator and compare it with other studies. Using a Rasch model, this study evaluated whether there are items that present difficulties in being answered or, conversely, that do not contribute to the measurement of depression. The Rasch analysis is based on a mathematical model that describes the relationship between the probability of a correct response to an item and the difference between the ability of the respondent. This model allows the joint measurement of people and items in the same construct. By doing this, a person's interpretation of scores is richer than just ranking from percentiles. The Survey on Health, Well-Being, and Aging in Latin America and the Caribbean (Project SABE) was used, a cross-sectional study that examines health conditions and functional limitations of persons aged 60 and older in the countries of Argentina, Barbados, Brazil, Chile, Cuba, Mexico, and Uruguay. This study also used the Costa Rican Longevity and Healthy Aging Study (CRELES), a set of nationally representative longitudinal surveys of health and life course experiences of older Costa Ricans. The scale shows good internal consistency. Cronbach's Alpha was 0.8 and more in most countries. Nevertheless, the Rasch model identified some items that do not fit the model. These items are: “Have you dropped or doing less of many of your activities and interests?”, “Were you afraid that something bad was going to happened to you?”, “Did you prefer to stay at home instead of going out and doing new things?”, “Did you feel you have more problems with your memory that other people your age?”, “Did you feel full of energy?”, “Did you think that most people were better off than you?”. It is important to mention that these items also present drafting problems. In the case of the item “Did you prefer to stay at home instead of going out and doing things?”, it can be thought that in the case of some Latin American countries, the preference for staying at home could be more related to the social and political context in which you live with depression. On the other hand, these results also invite us to think about the existence of various dimensions on this scale. Lack of fit is related to cultural contexts or to the fact that the data refers to a general population. Social, demographic, cultural, educational factors, gender roles, and social support networks could be evaluated in a multilevel model.
Resumen Con el cambio en los patrones demográficos y las mejoras en los sistemas de salud, se espera un aumento de la población adulta mayor en el mundo y especialmente en América Latina. Esto genera un aumento en la prevalencia de enfermedades crónicas y degenerativas; una de ellas es la depresión. La depresión supone costos sociales y financieros que pueden causar trastornos a la salud y a la vida cotidiana de quien la padece y de quienes están alrededor, ya que conduce a una morbilidad y mortalidad altas. El presente estudio pretende proporcionar más evidencia psicométrica sobre la idoneidad dela Escala corta de Yessavage (GDS-15) como instrumento para identificar síntomas depresivos en varias poblaciones de adultos mayores, sin patologías específicas, de diferentes países de Latinoamérica. Se utilizaron los datos de los estudios de la Encuesta sobre Salud, Bienestar y Envejecimiento en América Latina y el Caribe (SABE) y Costa Rica, y el Estudio de Longevidad y Envejecimiento Saludable (CRELES). Se hizo uso de la teoría clásica de los tests y un modelo de Rasch. Se encontró que la escala presenta buenos indicadores de consistencia interna con Alphas de Cronbach superiores a .8 en la mayoría de los países. Cuando se evalúa con el modelo de Rasch se identifican varios ítems que no ajustan al modelo, lo que podría relacionarse con los contextos culturales o al hecho de ser población general, por lo que se sugiere que se evalúe el efecto de factores sociales, demográficos, culturales, educativos, roles de género y red de apoyo social en un modelo multinivel.
