Método.

Participantes

Para este estudio se utilizan datos procedentes de un diseño en dos paneles con medidas repetidas de una muestra comunitaria de población adulta general. Para el primer panel se seleccionaron 1051 sujetos de ambos sexos y edades de 18 a 80 años, a partir de un conjunto de vecindarios representativos en términos sociodemográficos de un área urbana de aproximadamente un millón de habitantes (Valencia, España). De estos 1051 participantes, 740 proporcionaron datos completos a las medidas del estudio seis meses después.

Instrumentos

Versión breve (CESD-7). La versión de siete reactivos se configuró a partir de los resultados del trabajo de Santor y Coyne, en que se analizó una versión preliminar del CESD de nueve reactivos para estudiar la capacidad de esta escala para clasificar casos de depresión. Para comprobar la validez de constructo de la escala breve, en este trabajo se utilizan medidas de los ámbitos fisiológico (salud física) y social (integración social).

Resultados.

Los resultados mostraron que el CESD-7 puede ser descrito adecuadamente como unidimensional y que esta estructura factorial se mantiene invariante seis meses después. Además, la consistencia interna es aceptable y no varia sustancialmente en los dos paneles. A continuación se comprobó si la versión breve (CESD-7) se comportaba en forma equivalente a la versión original (CESD-20) en diferentes modelos de regresión lineal que buscaban analizar la validez de constructo del CESD al incorporar la dimensión temporal.

Discusión

Los resultados del estudio indican que la versión breve puede ser descrita como unidimensional y que esta estructura se mantiene invariante a los seis meses. Además, su fiabilidad (consistencia interna y fiabilidad test-retest) es adecuada. Al extender los análisis para comprobar la validez de constructo del CESD-7, se observó una gran similitud con la versión original (CESD-20). Esto sugiere que las puntuaciones del CESD-7 pueden utilizarse como un sustituto de la versión original de 20 reactivos sin perder por ello validez.

Method:

Participants

For this study we used data from a two-wave panel design with repeated measures in a community sample of Spanish adult population. In the first panel, 1051 participants of both sexes with ages ranging from 18 to 80 years completed the questionnaires. Participants in this panel were selected from a cross-section of representative neighbourhoods from a one million metropolitan area (Valencia, Spain). Participants completed questionnaires refering to their mental health, physical health, and social integration. Age, sex, educational level, and household income of participants were also coded.

After six months, almost 75% (N = 740) of the respondents completed the same questionnaires in the second panel. Attrition analyses between respondents and drop-outs showed nonsignificant differences in socio-demographic variables. A 54% of the respondents were women. The mean age was 39 years. Average educational level was high school (full-time education until 18 years) and average household income was 21500 euros (26000 US dollars, approximately).

Instruments

Original version of the CESD (CESD-20). The original Spanish version of the CESD-20 used in this paper was adapted by the authors in previous works. Previous research with the CESD-20 scores of the translated version showed a high degree of internal consistency and construct validity.

Brief version of the CESD (CESD-7). Based on the available literature, we selected seven items that showed the highest validity to classify cases of depression. Items for the revised measure included dysphoric mood (items 3, 6, and 18), motivation (item 7), concentration (item 5), loss of pleasure (item 16), and poor sleep (item 11 ).

As indicated above, we also measured health perceptions and social integration to further analyze the construct validity of this brief scale. Two instruments were selected:

Health Perception. We used the General Health Perception Questionaire developed by Davies and Ware to obtain information about the health status as appraised by the individual. The GHPQ includes 29 items with five category responses ranging from totally disagree to totally agree (e. g., "My health condition is excellent") that provides a global score with higher scores indicating better health perceptions.

Social Integration. To measure social integration we used the Social Integration in the Community Scale. This is a five-item scale that measures the sense of belonging and/or identification to a community or neighbourhood (e.g., "I feel identified with my community"). A higher score represents a higher level of social integration.

Results:

Results showed that the CESD-7 can be described as undidimensional and that this one-factor structure remains mostly invariant after six months. Internal consistency was adequate (α’ ≥ .82) in both panels. As for the validity of the brief version, we estimated several regression models for both the CESD-20 and CESD-7 as dependent variables. Predictors in these equations were: previous levels of depression, socio-demographic variables, physical health and social integration. Results showed a moderate relationship between measures of depression across time (CESD-20, β = 0.12, p < .001; CESD-7, β = 0.13, p < .001), and also that the CESD-20 and CESD-7 scores in panel two were significantly associated with sex, social integration, and physical health almost with the same strength for both versions. Also, non-significant associations were found for age, educational level, and household income for both versions. These results suggested that using the CESD-7 instead of the CESD-20 did not substantially change the results of linear regression models.

Discussion.

The results of this study indicate that the brief version of the CESD (CESD-7) has an adequate reliability and validity and that this brief measure is virtually equivalent to the original version (CESD-20) when used as a dependent variable in several linear regression models. Thus, both the original and brief versions scores were negatively and significantly associated with previous levels of good health conditions (perceived health) and social integration even after controlling for previous levels of depression in panel 1. There is extensive research showing that women report more depressive symptomatology than men. Also, physical health has been related with depression. Regarding social integration, there is also a vast array of empirical evidence relating it to the mental health of the individual. In our study, we found these expected associations both for the original CESD-20 and the brief version (CESD-7).

According to the results obtained in this study, we encourage researchers to use this brief measure of depression when survey space is limited or a fast and reliable measure of depression is needed.