Elsevier

Psychiatry Research

Volume 109, Issue 1, 31 January 2002, Pages 61-69
Psychiatry Research

Measurement adequacy of the CES-D among a sample of older African–Americans

https://doi.org/10.1016/S0165-1781(01)00360-2Get rights and content

Abstract

This manuscript examines the measurement adequacy of the CES-D when applied in older African–Americans (n=227). The response distribution was skewed towards less symptomatology on all items, and the internal consistency of the CES-D was acceptable (α=0.86). An exploratory factor analysis revealed four underlying factors of depressive symptomatology: (1) depressive/somatic; (2) positive; (3) interpersonal; and (4) social well-being. The primary factor was a combination of depression and somatic symptoms, a commonly reported uni-dimensional factor among ethnic groups. Additionally, a fourth factor emerged that has not previously been identified, labeled ‘social well-being’. This factor consisted of three items that may tap into the social interactions of the elderly: appetite, hopeful, and talk. These findings contribute to the growing body of evidence on the unique measurement properties of the CES-D across diverse populations.

Introduction

Survey instruments that assess mental health must measure actual symptomatology within individuals and groups, if prevention and treatment efforts targeting these symptoms are to be successful. However, real symptoms may be masked or over-stated when widely accepted survey instruments that have been validated among middle-aged whites are applied to minority populations across the life course. Social, cultural or language differences between population subgroups may result in differential understanding or response patterns that lead to non-equivalent measures of symptomatology. Furthermore, the application of ‘standard’ measures across age groups may lead to over- or underestimates of symptomatology if cohort differences affect response patterns on standard mental health surveys. The implications of such a situation include a misrepresentation of the attributes of minorities and the aged as well as possible distortion of the target, design and implementation of health promotion programs. Thus, as the US population becomes older and increasingly diverse, more information is needed on the appropriateness, reliability and validity of survey instruments for use among elderly minorities. This brief report investigates the measurement adequacy of the Center for Epidemiological Studies — Depression Scale (CES-D) among community-residing African–Americans aged 59+ (Radloff, 1977).

The CES-D is a commonly used survey instrument that measures depressive symptomatology in population-based surveys (Radloff, 1977). It also has been used as a screening tool for clinicians to identify persons who may need further evaluation for clinical depression, despite its original intent as a survey measure (Roberts et al., 1990, Zimmerman and Coryell, 1994, Lewinsohn et al., 1997, Lyness et al., 1997, Papassotiropoulos and Heun, 1999).

The CES-D was originally developed for the Community Mental Health Assessment, a population-based study of depressive symptoms (Radloff, 1977, Radloff and Teri, 1986). The 20-question survey contains items chosen from other validated mental health instruments designed to measure depressive symptoms. The questions measure symptoms experienced in the past week, using a scale from zero to three, with the highest possible score of 60 (representing more depressive symptoms).

The original four-factor structure included 16 of the 20 CES-D items: depressed affect (blues, depressed, lonely, cry, sad), positive affect (good, hopeful, happy, enjoy), somatic and retarded activity (bothered, appetite, effort, sleep, get going), and interpersonal feelings (unfriendly, dislike). The measure's psychometric properties were evaluated through studies conducted among three samples with successive repetition across several strata, including racial (black and white) and age subgroups (under 25, 25–64, and over 64). Although the factor loadings were reported exclusively for the three all white samples, the authors did note that the factor structure among the subgroups was similar (Radloff, 1977).

The measurement adequacy of the CES-D has come under considerable scrutiny since the original Radloff (1977) report. Several studies have found that the CES-D has acceptable internal consistency and a four-factor structure of depressive symptomatology (Roberts, 1980, Roberts et al., 1989, Blazer et al., 1998). Additionally, Mui et al. (2001) recently reviewed the literature on the CES-D and found that it is generally useful for assessing depression across diverse groups of older adults. However, there is growing evidence that the measurement adequacy of the instrument varies when applied to minority groups.

Roberts (1980) reported acceptable reliability, as well as the expected factor structure for both Mexican–Americans and African–Americans. Also, a study of Anglo–American and Mexican–American psychiatric patients supported the a priori four-factor model using confirmatory factor analysis (Roberts et al., 1989). The authors did not find an effect of language or ethnic status on the psychometric properties of the CES-D among these groups.

However, countervailing evidence of the CES-D's factor structure has led to questions regarding its applicability across different ethnic populations. Using confirmatory factor analysis, Chapleski et al. (1997) found the factor solution of the CES-D was inconsistent with the original four-factor structure when applied to American Indians. A shortened 12-item CES-D, previously proposed for use with Mexican–Americans (Liang et al., 1989), provided a better fit to their data on American Indians.

In addition, several studies have shown that affective and physical responses to depression may not represent unique constructs for some minority populations. Two studies of American Indians (Baron et al., 1990, Somervell et al., 1993), one study of Mexican–Americans (Garcia and Marks, 1989), and a study of Hispanics composed of Mexican–Americans, Puerto Ricans, and Cuban–Americans found a three-factor structure of depressive symptoms, which combined depressive and somatic complaints (Guarnaccia et al., 1989). Additionally, Kuo (1984) reported that Chinese–American performance on the CES-D was comparable to that of other ethnic groups, reflecting the three-factor structure with no distinct somatic factor.

In a study of the CES-D's psychometric properties among race and gender subgroups of older persons, white and African–American women demonstrated the expected four-factor structure using exploratory factory analysis, although two items (‘mind’ and ‘failure’) did not load onto any factor for African–American women (Callahan and Wolinsky, 1994). Furthermore, the authors found an obscure seven-factor structure among African–American men, including three new factors for anxiety, introspection and crying. Blazer et al. (1998) reported that the CES-D met minimal standards for the hypothesized factor structure for older whites and African–Americans using confirmatory factor analysis. Nonetheless, they showed a very high correlation (0.82) between depressed affect and somatic complaints, as well as a lack of inter-factor correlation between positive affect and other factors for the African–American sample.

In a study of item-response bias, African–Americans were significantly more likely than whites to respond affirmatively on two interpersonal questions, ‘people are unfriendly’ and ‘people dislike me’ (Cole et al., 2000). The authors of this study argue that racial differences in responding to these two items may reflect a sense of perceived racial prejudice among African–Americans that may compromise the validity of the interpersonal items for use with this population. They also found that the question, ‘crying spell’, was gender-biased, and thus proposed a shortened 17-item CES-D excluding all three of these items.

To date, the published reports examining the psychometric properties of the CES-D suggest differential factor structures across ethnic groups, despite its general acceptance and use among researchers and clinicians. Among the articles that have tested the measurement adequacy of the CES-D, there is evidence that the original four-factor structure may not be applicable to an increasingly diverse population.

In addition to questioning the acceptability of the CES-D among different ethnic groups, several mental health researchers have examined the appropriate use of the CES-D across the life course. Some have suggested that CES-D scores are inaccurately inflated among the elderly because older persons may experience a decline in physical health leading to higher scores on some of the CES-D items, especially somatic complaints. Williamson and Schulz (1992) found that poor subjective and objective health leads to higher CES-D scores among community-dwelling older adults, although older persons are at no greater risk of depression after controlling for frailty (Lewinsohn et al., 1991, Blazer et al., 1991, Andresen et al., 1994). Additionally, the CES-D has been shown to be unbiased by somatic complaints of elderly subjects (Radloff and Teri, 1986, Foelker and Shewchuk, 1992).

Kessler et al. (1992) found a non-linear relationship between somatic and depressive affect scales of the CES-D among persons aged 60 and older, suggesting that depressive symptoms may vary over the life course, rather than a confounding relationship between depressive symptoms and age-related somatic complaints. As previously described, Liang et al. (1989) found that an a priori 12-item factor model fit the data well within three generations of Mexican–Americans. However, the study also showed structural variations across the generations that could not be explained by socioeconomic characteristics, acculturation or physical health.

Based on the literature of race, age, and the CES-D, additional research is warranted to examine the measurement adequacy of the CES-D across racially diverse groups, especially within age-specific groups. Thus, this study provides a basic psychometric analysis of this standard instrument of depressive symptoms among a sample of older African–Americans.

Section snippets

Study participants

To examine the measurement adequacy of the CES-D for use in surveys of older African–Americans, we used data from the Durham Elders Project, a study sponsored by the Center on Minority Aging (CMA) at the University of North Carolina at Chapel Hill. The CMA is one of six sites funded by the National Institute on Aging and the National Institute of Nursing Research and the Office of Research on Minority Health as a Resource Center on Minority Aging Research (RCMAR). One of the primary goals of

Description of the sample

The individuals were an average of 74 years old (range 59–96) at the time of this study and predominantly female (75%). Approximately 38% of the respondents were married, 47% were widowed, and 11% were divorced or separated. Income, measured categorically with 14 response options, ranged from under $3000 to over $100 000; the median income was $10 000–$14 999 (n=206 respondents provided income data). Eighty percent of the sample reported their health to be good or excellent on a standard

Discussion

The descriptive results of the CES-D were similar to previous reports among this voluntary sample of older African–Americans. Responses were skewed towards less symptomatology, and 14% of the sample scored 16 or higher on the CES-D score, comparable to other reports of older African–Americans (Callahan and Wolinsky, 1994) and older persons in general (Lewinsohn et al., 1997).

The factor analysis provides additional support for the growing body of evidence that the CES-D does not have one

Acknowledgements

This work was conducted at the Center on Minority Aging at the University of North Carolina at Chapel Hill, 730 Airport Rd, CB# 3465, Chapel Hill, NC 27599, USA.

References (26)

  • D.G. Blazer et al.

    Symptoms of depression among community-dwelling elderly African–American and White older adults

    Psychological Medicine

    (1998)
  • C.M. Callahan et al.

    The effects of gender and race on the measurement properties of the CES-D in older adults

    Medical Care

    (1994)
  • E.E. Chapleski et al.

    Structure of a depression measure among American Indian elders: confirmatory factor analysis of the CES-D Scale

    Research on Aging

    (1997)
  • Cited by (0)

    View full text