Oliver D, Hopper A, Seed P
Academic Department of Elderly Care, Guy's Kings and St. Thomas' School of Medicine and Dentistry, London, England.
J Am Geriatr Soc. 2000 Dec;48(12):1679-89. doi: 10.1111/j.1532-5415.2000.tb03883.x.
To analyze published hospital fall prevention programs to determine whether there is any effect on fall rates. To review the methodological quality of those programs and the range of interventions used. To provide directions for further research.
Systematic review of published hospital fall prevention programs. Meta-analysis.
Keyword searches of Medline, CINAHL, monographs, and secondary references. All papers were included that described fall rates before and during intervention. Risk ratios and 95% Confidence Intervals (95% CI) were estimated and random effects meta-analysis employed. Begg's test was applied to detect possible publication bias. Separate meta-analysis regressions were performed to determine whether individual components of multifaceted interventions were effective.
A total of 21 papers met the criteria (18 from North America), although only 10 contained sufficient data to allow calculation of confidence intervals. A rate ratio of <1 indicates a reduction in the fall rate, resulting from an intervention. Three were randomized controlled trials (pooled rate ratio 1.0 (CI 0.60, 1.68)), seven prospective studies with historical control (0.76 (CI 0.65, 0.88)). Pooled effect rate ratio from these 10 studies was 0.79 (CI 0.69, 0.89). The remaining 11 studies were prospective studies with historical control describing fall rates only. Individual components of interventions showed no significant benefit.
The pooled effect of about 25% reduction in the fall rate may be a result of intervention but may also be biased by studies that used historical controls not allowing for historical trends in the fall rate before and during the intervention. The randomized controlled trials apparent lack of effect might be due to a change in practice when patients and controls were in the same unit at the same time during a study. Studies did not analyze compliance with the intervention or opportunity costs resulting from the intervention. Research and clinical programs in hospital fall prevention should pay more attention to study design and the nature of interventions.
分析已发表的医院跌倒预防项目，以确定其对跌倒率是否有影响。评估这些项目的方法学质量以及所采用干预措施的范围。为进一步研究提供方向。
对已发表的医院跌倒预防项目进行系统评价。荟萃分析。
通过检索Medline、CINAHL、专著及二次参考文献。纳入所有描述干预前和干预期间跌倒率的论文。估计风险比和95%置信区间（95%CI），并采用随机效应荟萃分析。应用Begg检验检测可能的发表偏倚。进行单独的荟萃分析回归，以确定多方面干预措施的各个组成部分是否有效。
共有21篇论文符合标准（18篇来自北美），但只有10篇包含足够数据以计算置信区间。率比<1表明干预导致跌倒率降低。三项为随机对照试验（合并率比1.0（CI 0.60，1.68）），七项为有历史对照的前瞻性研究（0.76（CI 0.65，0.88））。这10项研究的合并效应率比为0.79（CI 0.69，0.89）。其余11项研究为有历史对照的前瞻性研究，仅描述了跌倒率。干预措施的各个组成部分未显示出显著益处。
跌倒率约降低25%的合并效应可能是干预的结果，但也可能因使用历史对照的研究存在偏倚，这些研究未考虑干预前和干预期间跌倒率的历史趋势。随机对照试验明显缺乏效果可能是由于在研究期间患者和对照在同一病房时实践发生了变化。研究未分析干预措施的依从性或干预带来的机会成本。医院跌倒预防的研究和临床项目应更加关注研究设计和干预措施的性质。
