Dewey M E, Saz P
Trent Institute for Health Services Research, Medical School, University Hospital, Nottingham, NG7 2UH, UK.
Int J Geriatr Psychiatry. 2001 Aug;16(8):751-61. doi: 10.1002/gps.397.
No recent attempt has been made to synthesise information on mortality and dementia despite the theoretical and practical interest in the topic. Our objective was to estimate the influence on mortality of cognitive impairment and dementia.
Data sources were Medline, Embase, personal files and colleagues' records. Studies were considered if they included a majority of persons aged 65 and over at baseline either drawn from a total community sample or drawn from a random sample from the community. Samples from health care facilities were excluded. The search located 68 community studies. Effect sizes were extracted from the studies and if they were not included in the published studies, effect sizes were calculated where possible: this was possible for 23 studies of cognitive impairment and 32 of dementia. No attempt was made to contact authors for missing data.
For the studies of cognitive impairment Fisher's method (a vote counting method), gave a p-value (from eight studies) of 0.00001. For studies of dementia, age-adjusted confidence intervals (CI) were pooled (odds ratio (OR) 2.63 with 95% CI 2.17 to 3.21 from six studies).
Levels of cognitive impairment commonly found in community studies give rise to an increased risk of mortality, and this appears to be true even for quite mild levels of impairment. The analysis confirms the increased risk of mortality for dementia, but reveals a dearth of information on the causes of the excess mortality and on possible effect modification by age, dementia subtype or other variables.
尽管该主题在理论和实践方面都备受关注，但近期尚无对死亡率和痴呆症相关信息进行综合分析的尝试。我们的目标是评估认知障碍和痴呆症对死亡率的影响。
数据来源包括医学期刊数据库（Medline）、荷兰医学文摘数据库（Embase）、个人档案以及同事的记录。若研究纳入的基线人群中大多数为65岁及以上老年人，且这些人群来自全社区样本或社区随机样本，则该研究被纳入考虑范围。来自医疗机构的样本被排除在外。通过检索共找到68项社区研究。从这些研究中提取效应量，若效应量未在已发表的研究中给出，则尽可能进行计算：23项认知障碍研究和32项痴呆症研究的效应量可进行计算。未尝试联系作者获取缺失数据。
对于认知障碍研究，费舍尔方法（一种投票计数法）得出的p值（来自八项研究）为0.00001。对于痴呆症研究，对年龄调整后的置信区间进行合并（六项研究得出的比值比（OR）为2.63，95%置信区间为2.17至3.21）。
社区研究中常见的认知障碍水平会导致死亡率上升，即便在相当轻度的障碍水平下似乎也是如此。该分析证实了痴呆症患者死亡率上升这一情况，但揭示了关于死亡率过高的原因以及年龄、痴呆症亚型或其他变量可能产生的效应修正方面信息的匮乏。
