Alterations in cognitive function during the menopausal transition.

Abstract

To the Editor: The physiological and psychological changes associated with menopause have remained a major psychosocial stressor for women throughout the ages, but many of the neuropsychological changes have not been well characterized. The present study evaluates cognitive function in a large group of women aged 40 to 54 who were enrolled in the Kinmen Women-Health Investigation (KIWI).1 Kinmen is a 176-km2 island located 154 miles (248 km) west of Taiwan and 25 miles (41 km) east of mainland China. It consists of four townships with a total population of 51,060 (1998). The people in Kinmen are Han Chinese, and most are involved in farming activities. The details of the KIWI have been described elsewhere.1 Of the 1,270 subjects in the final analysis, 77 (6%) were current or past hormone replacement therapy (HRT) users. The 1,173 women who never used HRT had a significantly lower education level than HRT users (P = .004). The distribution of menopausal status of women who never used HRT was premenopausal, 694 (58%); perimenopausal, 323 (27%); and postmenopausal, 176 (15%). The premenopausal period was defined as regular menstruation. A woman was considered perimenopausal if her menstrual cycles had been irregular or her last menstrual bleeding occurred more than 3 and 12 or fewer months before the study. Women who had not menstruated within the previous 12 months were categorized as postmenopausal. Women with surgically induced menopause were excluded from this study. Each subject received a 45-minute battery of neuropsychological tests consisting of: the Wechsler Adult Intelligence Scale-Revised (WAIS-R),2 Digit Span Subtest (forward and backward), the continuous recognition paradigm of Kimura (CRP),3 the Trail Making Test (TMT) parts A and B,4 verbal fluency,5 and the Rey Auditory-Verbal Learning Test (RAVLT).6 The menopause-related symptom checklist modified from the Kupperman index7 and the Hospital Anxiety and Depression Scale (HADS)8, 9 were also administered to each subject. Table 1 lists the results of the neuropsychological tests by menopausal status group for the non-HRT users. Univariately, all cognitive functions except the CRP significantly declined with the progression of menopausal status (P < .0001, except for verbal fluency P = .008). Pairwise comparisons further suggested that most of the differences derived from a comparison between the postmenopausal group and the other two groups. The univariate analysis also indicated that age and education were two important factors that were significantly associated with cognitive function. However, the HADS scores did not demonstrate an association with any of the cognition tests. Similar results were obtained in the multivariate analysis. With menopausal status, age, education, and HADS score in the model, education and age were significantly related to most cognitive functions. Education was a significant covariate for all cognition tests (P < .0001) except for the CRP (P = .12). The higher the education level, the better the cognitive function. Age was a significant covariate for the backward digit span (P < .001), CRP (P = .004), and TMT part B (P < .0001). Cognitive function declined significantly with age. Of all neuropsychological tests, education was the only factor that influenced the forward digit span, TMT part A, verbal fluency, and RAVLT, based on the results of the linear model. The backward digit span and TMT part B results were associated with age and education. After adjusting for age and education, menopausal status had no effect on cognitive function except for the TMT part A. No significance was found for the effect of the HADS score on any cognitive function. Multivariate analysis of HRT effects in perimenopausal and postmenopausal women showed that, after adjusting for menopausal status, age, education, presence of menopausal symptoms, HADS score, and HRT usage (never used, used <6 months, used ≥6 months), HRT only affected backward digit span (P = .001). Those using HRT for at least 6 months had the highest backward digit span score, followed by those using HRT for less than 6 months and then by the nonusers. The overall significant difference was attributed to the pairwise difference between nonusers and those who used HRT for less than 6 months; the adjusted relative risk was 1.38 (1.15–1.67). In this community sample population, we found that most of the cognitive functions had significantly decreased from the pre- to the postmenopausal stage. The difference disappeared after adjusting for age and education, except for the TMT part A. We also demonstrated that HRT might help in the performance of the backward digit span of the WAIS-R.