Prevalence of cognitive impairment in an urban Hispanic community population.

Abstract

To the Editor: Dementia is a major health problem in older persons, but few studies have addressed the prevalence of dementia specifically among Hispanics in the United States. The purpose of this study was to estimate the prevalence of cognitive impairment in a community-living sample of older urban Hispanics living in Chicago using the Mini-Mental State Examination (MMSE). A convenience sample (N = 222) of self-identified Hispanics aged 55 and older was administered cognitive testing at nine community sites throughout Chicago by two bilingual medical students. The full (30-point) MMSE was administered in the subject's preferred language, English (16.6%) or Spanish (83.3%). This version uses “apple,”“table,” and “penny” as the three words for recall. Serial sevens was not used, but spelling “world” backwards was. The phrase “no ifs, ands or buts” was translated to “ni si, ni no, ni peros.”1 Two hundred twenty-two subjects were enrolled, predominantly of Mexican origin (48.6%), Mexican ancestry born in the United States (20.3%), or Puerto Rican origin (19.4%), which closely reflects the proportions of Hispanics in Chicago. There were 38 men (17.1%) and 184 women (82.9%). Mean age ± standard deviation of the overall sample was 70.3 ± 7.6 (range 55–92), and mean education was 5.6 ± 4.5 years (range 0–18). The prevalence of cognitive impairment (MMSE score <24) was 26.6%. The prevalence of moderate cognitive impairment (MMSE score <18) was 10.4%. After adjusting for age and education, 12.6% of the sample had an adjusted MMSE score of less than 24.2 Mean MMSE score declined with age, from a mean of 26.9 in those aged 55 to 64 to 18.7 in those aged 85 and older. Mean MMSE scores increased with years of schooling from 20.9 in those with 0 to 1 years of education to 27.8 in those with 12 or more years of education. Mean MMSE scores differed significantly between English- (28.1 ± 2.3) and Spanish-speaking (24.5 ± 5.5) subjects (P < .001). Mean MMSE scores also differed significantly between immigrants (24.5 ± 5.4) and U.S.-born subjects (27.3 ± 3.9) (P < .001). Factors that did not significantly affect mean MMSE included gender (P < .418), diabetes mellitus (P < .633), hypertension (P < .8), and smoking (P < .48). The prevalence of cognitive impairment in this urban, community-living Hispanic population was much higher than the average reported for the United States. Cognitive impairment increased with age and decreased with education. The impairment in this study may reflect a truly higher prevalence of dementia, or it may be an overestimate resulting from sampling, cultural, and test biases. A higher prevalence of dementia is plausible given the relatively high rates of related health problems in Hispanics that are risk factors for vascular dementia. A self-reported history of these disorders was indeed high in this study (hypertension, 53.3%; diabetes mellitus, 33.3%; cigarette smoking, 22.5%) (see Appendix 1). However, the higher prevalence of dementia may be partially due to a bias of the MMSE for disadvantaged minority populations. Bird et al. and Escobar et al. found that education significantly influenced MMSE score.3,4 In this study sample, 61.7% had less than 7 years of education, resulting in low levels of literacy and limited experience with test taking. Investigators have proposed various corrections for the educational and cultural biases of the MMSE. Murden et al. postulated that a score of less than 18 in those with low education was an indicator of only mild cognitive impairment.5 Mungas et al. calculated a statistical correction for the bias of education and age on MMSE scores in a population composed of Hispanics and non-Hispanics.2 Escobar et al. and Teresi et al. suggested differentially weighing or removing five biased questions, equal to nine points.4,6 Recently, many authors have argued that the MMSE score reflects real cognitive impairment secondary to educational cognitive factors.7 Significant differences in mean MMSE scores between English- and Spanish-speaking subjects and between immigrants and U.S.-born subjects in this study suggest that ethnic and language biases may falsely lower the mean total MMSE score. Therefore, the need for a correction in MMSE score, whether by modification of the cutoff, statistical adjustment, or removal of biased items, is likely appropriate to prevent an artificially high reported prevalence of cognitive impairment in groups such as this sample studied. Other limitations of this study include convenience sampling bias and the lack of an unequivocal diagnosis of dementia. The point prevalence of cognitive impairment (MMSE <18) in older Chicago Hispanics probably reflects a truly high prevalence of dementia despite cultural and educational biases of the MMSE test. More resources need to be directed toward the study of dementia in the rapidly growing, underserved Hispanic population.