目標：在高齡化人口中，伴隨老化而來的認知障礙益發受到重視。如何維持中老年人認知功能是一重要課題。本研究欲藉國內大型世代資料，分析台灣中老年人閱讀與認知衰退的關聯性。方法：本研究分析「台灣地區中老年身心社會生活狀況長期追蹤調查」2003及2007年資料。以2003年為基礎點，排除有認知功能缺損或不識字者，及非本人作答或資料不全者，得3,147位≥50歲之研究個案。以羅吉斯迴歸(logistic regression)控制社會人口學、健康狀況及健康行為等變項，分析高低閱讀頻率與四年後認知衰退之關聯。結果：若將閱讀活動保留在所有休閒活動(共十項)中，結果顯示高休閒參與者(≥3項/週)的認知衰退風險較低(OR=0.56, p<0.001)。若將閱讀活動從休閒項目中個別獨立出來，結果顯示高休閒參與者的認知衰退風險雖仍較低(OR=0.73, p=0.050)，但關聯性較弱；而有高閱讀頻率者的衰退風險則明顯低於低閱讀頻率者(OR=0.51, p<0.001)。另依教育程度分群分析，發現高休閒參與或高閱讀頻率(OR=0.57, 0.50; p<0.01)皆對低受教育年數(0-6年)受訪者的認知衰退有延緩作用，但對高教育年數受訪者的影響則不明顯。結論：本研究凸顯閱讀對預防中老年人認知退的重要性。建議中老年人養成每週至少閱讀一次的習慣，以促進認知功能健康及延緩衰退。

OBJECTIVES: Cognitive impairment in the aging process has become increasingly critical, especially in aging populations. Thus maintaining the cognitive function of older adults is crucial. The present study aimed to determine the association between reading and cognitive decline in older Taiwanese people by using national cohort survey data. METHODS: The data sets of the Taiwan Longitudinal Survey on Aging from 2003 and 2007 were analyzed. After we excluded those who were cognitively impaired or unable to read in 2003 (baseline), in addition to proxy-reported or incomplete data, a total of 3,147 eligible participants ≥50 years old remained. The logistic regression model was used to determine the association between reading and the subsequent 4-year cognitive decline, adjusted for sociodemographic variables, health status, and health behaviors. RESULTS: When reading was included in the 10 types of leisure activities, our results showed that the likelihood of cognitive decline was lower for participants with a higher level (≥3 types/week) of leisure activities (OR=0.56, p<0.001). When reading was excluded from the leisure activities, the results still suggested a lower likelihood of cognitive decline for participants with a higher level of leisure activities (OR= 0.73, p=0.050); however, the strength of the association was apparently lower. Nevertheless, participants exhibiting a higher frequency of reading were less likely to suffer from cognitive decline compared with those having a lower frequency of reading (OR=0.51, p<0.001). Furthermore, our results of subgroup analysis showed that a higher level of leisure activities or reading (OR=0.57, 0.50; p<0.01) could delay cognitive decline of participants with a lower educational level (0-6 years of formal education), while the effects were not significant for those with higher educational levels. CONCLUSIONS: Our findings demonstrate the advantage of reading to delay cognitive decline in older adults. We suggest that older adults read once a week or more to promote health and delay cognitive decline.