ベイズ推計メタアナリシス：75歳超でのスタチン投与ベネフィット

事前確率推定した上でのベイズ推計手法でみると、各種リスク要素を具有する確率の多い75歳超でのスタチン投与の真の有益性がわかるのではないか？





Meta-analysis of Usefulness of Treatment of Hypercholesterolemia with Statins For Primary Prevention in Patients Older Than 75 Years
John B. Kostis, et al.
The American Journal of Cardiology
Published:January 27, 2020
DOI:https://doi.org/10.1016/j.amjcard.2020.01.020
https://www.ajconline.org/article/S0002-9149(20)30061-8/fulltext




American College of Cardiology (ACC)/American Heart Association (AHA) guidelineでは75歳超での一次予防のためのスタチン使用を推奨していない。" 2019 Task Force for the management of dyslipidemias of the European Society of Cardiology (ESC) and European Atherosclerosis Society (EAS)"では IIb推奨として、高リスクなら75歳超での一次予防のためのスタチン使用開始が推奨される。
75歳超対象に、ベイズ解析にて改めて35のRCTを解析


Methodの一部

Twenty one of the 35 studies included data on persons ＞75 years as well as ＜75 and 14 studies had only data on persons ＜75.
The studies were analyzed using both the frequentist and Bayesian approaches. For all-cause death, the log of the risk ratio (treatment/placebo) was modeled by a normal distribution.　The prior distribution was assumed to have a mean equal to a weighted average of the logarithm of risk ratio (RR) with weights equal to the inverse of the sum of the estimates of between study variability and within study variability plus a drift term.　 The prior variance was divided into two components, an estimate of between-study variability and an estimate of within-study variability.
 The posterior distribution was calculated using Bayes’ rule and the 95% confidence interval was determined using the highest posterior density region. 

In toto, 192,079 individuals and 18,260 deaths were included in this analysis and were subdivided into 2 groups: patients aged 75 or younger versus patients older than 75. Data were extracted by two reviewers (JBK, MG) independently and disagreements were adjudicated by face to face discussion.

米国および欧州の臨床ガイドラインでは、75歳以上の高コレステロール血症患者の初期予防のためにスタチンによる治療を推奨していません。 一次予防のためのスタチン療法をプラセボまたは通常のケアと比較したこの年齢層の35件のランダム化比較試験のデータを分析しました。 結果として全死因を使用して、2種類の分析を実行しました： frequentist とBayesian。
Frequentist analysis＜頻度確率解析＞では、症例（スタチン）と対照（プラセボまたは通常のケア、p = 0.16）の死亡率に有意な差は示されませんでした。
しかし、ベイジアン分析では、75歳以上の人はスタチン治療による死亡率が低かった（p = 0.03）。
結論として、ベイジアン分析は、75歳以上の人における一次予防のためのスタチン治療の明確で統計的に有意で臨床的に関連するベネフィットを示しています。

For all-cause death frequentist analysis indicated a non-significant effect for treating patients older than 75 with statins. However, using Bayesian analysis,a significant benefit in terms of all-cause death is observed.
Redline is the prior distribution
Blackline shows data in studies on patients older than 75
Greenline is posterior distribution after consideration of the prior distribution and the data available on patients older than 75 using Bayes’rule