Response to the commentary letter entitled 'Diagnostic radiography and thyroid cancer - causation or simply an association?' to our article entitled 'Diagnostic radiography exposure increases the risk for thyroid microcarcinoma: a population-based case-co
Response to the commentary letter entitled 'Diagnostic radiography and thyroid cancer - causation or simply an association?' to our article entitled 'Diagnostic radiography exposure increases the risk for thyroid microcarcinoma: a population-based case-co
复制标题
对题为“放射诊断与甲状腺癌——因果关系还是简单的关联？”的评论信的回应
DOI：
10.1097/cej.0000000000000206
复制
发表时间：
2016
期刊：
影响因子：
--
通讯作者：
Udelsman,Robert
中科院分区：
文献类型：
--
作者：
Zhang,Yawei;Huang,Huang;Sandler,Jason;Udelsman,Robert
在病例对照研究中，如果在研究实施过程中没有运用适当的策略，回忆偏倚可能会对研究结果产生显著影响。在我们的研究中，访谈者对研究假设不知情，问卷还包括其他暴露问题，如生活方式，病史，职业和饮食。我们重视并赞赏Wiltshire博士和Balasubramanian博士的建议，即在问卷中包括“随机/不相关”的问题，如关于超声和MRI扫描的问题，将有助于估计回忆偏差的程度，以便于未来的研究。尽管回忆偏差本身无法量化，但由于缺乏对暴露与结果之间“真实”关联的了解，通常可以辨别出偏置的方向。在我们的研究中，阳性结果仅限于甲状腺微小癌，这支持了差异回忆偏差在我们的研究中不太可能发挥主要作用的观点。因此，潜在的回忆偏倚可能是无差别的，这使结果偏向零。
Recall bias could have a significant impact on the results generated from a case–control study if there were no precautious strategies being exercised during the implementation of the study. In our study, the interviewers were blinded to the study hypothesis, and the questionnaire also included questions on other exposures such as lifestyle, medical history, occupation, and diet. We value and appreciate the suggestion by Dr Wiltshire and Dr Balasubramanian that including ‘random/unrelated’questions in the questionnaire, such as questions about ultrasound and MRI scans, would be helpful in estimating the extent of recall bias for the sake of future studies.Although the recall bias cannot be quantified per se, because of the lack of knowledge about the ‘true’association between exposure and outcome, the direction of bias can often be discerned. In our study, the positive findings were only limited to thyroid microcarcinoma, supporting the notion that differential recall bias was less likely to play a major role in our study. Therefore, the potential recall bias was likely to be nondifferential, which biased the results toward null.