In Reply

Abstract

Editors note: The above letter was referred to the authors of the original paper, and their reply follows. In reply: We appreciate the invitation to respond to Rozzini et al.'s letter. In a previous study,1 the estimated survival rates during 42 months, obtained with the Kaplan-Meier method, tended to decrease with a decline in intellectual functioning as measured by an assessment instrument developed by the Social Survey Division of the Office of Population Censuses (OPCS)2 in Great Britain among Japanese community-residing older people. Application of the Cox proportional hazards model resulted in unadjusted hazard ratios (HRs) of mild, moderate, and severe intellectual dysfunctioning for mortality of 1.68, 2.44, and 5.37, respectively. Multivariate analysis, on the other hand, yielded adjusted HRs of mild, moderate, and severe intellectual dysfunctioning of 1.19, 1.12, and 1.74, respectively, when adjustments were made for age, sex, general health status, health management, and psychosocial conditions, leaving severe intellectual dysfunctioning as the only significant predictive factor. However, some important factors, such as education level or depressed mood, were not assessed in our study for each of the older subjects as a factor correlated with cognitive performance.3,4 Thus, our study had limitations in its understanding of the relationship between intellectual dysfunctioning and mortality. In response to our cautions about the potential effect of education levels or depressed mood on intellectual functioning, Rozzini and colleagues have examined the relationship between cognitive status measured with the Mini- Mental Status Examination (MMSE) and mortality among community-residing older people in Italy, controlling for length of education and for depressed mood, measured with the Geriatric Depression Scale (GDS), as confounding factors. They demonstrated that the estimated survival rates during 60 months clearly decreased with a decline in MMSE levels. Furthermore, the firm linear increase of unadjusted or adjusted HRs was demonstrated through the groups of decreasing MMSE levels, and the adjusted HRs for mortality showed that older people with a MMSE score of 25–27, a MMSE score of 19–24, and a MMSE score >19 were estimated to be, respectively, 1.2, 1.4, and 2.9 times as likely as those with a MMSE score < 27 to die within 60 months. These findings underline clearly the dose-response association between intellectual dysfunctioning and mortality among community-residing older people. We are encouraged that Rozzini and colleagues were able to confirm a strong prediction value of mortality of cognitive function for community-residing older people, independent of education level and depressed mood, and to extend our work in community-residing older people in Italy.