目的·方法.検診の効果とバイアスについて,これまでの知見を整理して解説する.結果·結論.検診の効果については,対象とするがんの死亡率を評価指標とするのが一般的である.検診発見例における病期分布や生存率を症状発見例と比較することは,一連の検診評価プロセスの中で重要な評価指標ではあるが,種々のバイアス(selfselection bias,lead time bias,length bias,overdiagnosis bias)の影響を受ける可能性が高く,死亡減少効果に代わりうるものではない.Overiagnosisは,バイアスとしての意義に加えて,検診による不利益としての意義が大きくなりつっある.死亡率減少効果の評価方法としての国際水準はランダム化比較試験(Randomized Controlled Trial)であり,諸外国ではRCT以外の研究デザインは証拠として軽視される傾向にある.しかし,RCTの中でも計画·実行·解析が適切にされたかの吟味が重要であり,効果に関する証拠のまとめを作成する際には,研究デザインだけでなく,研究の質のチェックが手順の中に組み込まれつつある.Objective and Methods. This aims to overview the effect of screening and related biases by reviewing previous findings. Results and Conclusion. The effect of screening is usually measured by site-specific mortality rate as an endpoint. Although comparisons of stage distribution or survival between screen-detected and symptom-detected cases are important measures in the entire process of screening evaluation, they will not be surrogates for mortality due to high possibility of suffering from various biases, such as self-selection bias, lead time bias, length bias. In particular, overdiagnosis has became of great concern not only as bias but also as harm due to screening.For evaluating mortality reduction, randomized controlled trial is a gold standard method. In the US and European countries, the results from any study design other than RCT, tend to be considered less valuable. However, it is important to evaluate how well the study designed, conducted and analyzed even for RCT. In the process of summarizing the evidence, checking internal validity is now concerned essential part of ranking the quality of study results.
展开▼