孩童時期接受電腦斷層與日後產生癌症風險之統合分析

Childhood CT Scans and Subsequent Cancer Risk-A Meta-Analysis

目的：蒐集過去的觀察型研究，以統合性分析方法分析童年時期的電腦斷層暴露，與之後發生癌症的相關性；並藉由subgroup analysis，探討不同研究之間可能存在的bias和異質性(heterogeneity)。方法：運用Pubmed關鍵字搜尋以及參考資料研讀蒐集相關文獻，計算Summary relative risk。並用固定式模型(fixed-effects model)及隨機式模型(random-effects model)進行分析。另外，進一步使用分組統合分析(subgroup analysis)探討tumor-predisposing factors與reverse causation對結果的影響。結果：我們最終蒐集到六篇retrospective cohort studies和一篇case-control study。綜合所有種類的癌症，summary RR為1.20(95% CI 1.11 to 1.29)。若排除具有tumor-predisposing factors的個案，summary RR增加為1.31(1.02 to 1.68)。當不分部位，綜合所有CT當作exposure，產生Leukemia的summary RR為1.24(0.99 to 1.55)；產生brain tumor的summary RR為1.47(1.11 to 1.94)。分組統合分析(subgroup analysis)顯示若研究使用較長的latency period，得到的RR會比較低(1-year RR= 2.42; 2-year RR=1.25; 5-year RR=1.97)。The funnel plot結果不對稱，代表可能有出版性偏差存在。結論：本研究之結果顯示孩童時期接受電腦斷層可能會增加日後產生癌症之風險，尤其是腦瘤。然而，由於存在出版性偏差，需要有更多後續的研究確認此因果關係。

Objectives. To review the published literature to evaluate the association between the performance of CT scans in childhood and subsequent tumor risk and to assess potential sources of bias and heterogeneity in those studies.
Methods. Studies were identified by searching Pubmed and bibliographies of relevant articles. Summary relative risks (RRs) with 95% confidence intervals (CIs) were calculated using fixed and random effects models as well as the Shore method. Subgroup analyses were performed to explore the impact of tumor-predisposing factors and reverse causation.
Results. Six retrospective cohort studies and one case-control study were analyzed. When summarizing data from studies in which different types of cancer were combined, we found a summary RR of 1.20 (95% CI 1.11 to 1.29; 7 studies). After excluding results that involved subjects with tumor-predisposing factors, the summary RR increased to 1.31 (1.02 to 1.68; 5 studies). When summarizing data from studies in which all types of CT scans were combined as exposure, we found a summary RR of 1.24 (0.99 to 1.55; 6 studies) for leukemia and an RR of 1.47 (1.11 to 1.94; 5 studies) for brain tumors. Subgroup analyses on latency periods revealed that a longer latency period seemed to decrease the RRs (1-year RR= 2.42; 2-year RR=1.25; 5-year RR=1.97). Funnel plot showed evidence of asymmetry, indicating possible publication bias.
Conclusion. The findings of this study suggest an association between the performance of CT scans in childhood and increase risk of tumor, especially brain tumors. However, because there was possibly publication bias, there is a need for further studies to validate the positive association we found.