Reply: CDC analysis of ICSI/autism: association is not causation.
Reply: CDC analysis of ICSI/autism: association is not causation.
复制标题
回复：CDC 对 ICSI/自闭症的分析：关联不是因果关系。
DOI：
10.1093/humrep/dev107
复制
发表时间：
2015
期刊：
影响因子：
--
通讯作者：
Jamieson,DeniseJ
中科院分区：
文献类型：
--
作者：
Kissin,DmitryM;Zhang,Yujia;Boulet,ShereeL;Fountain,Christine;Bearman,Peter;Schieve,Laura;Yeargin-Allsopp,Marshalyn;Jamieson,DeniseJ
Barad博士等人表示担心，该分析没有根据“种族/民族和/或社会经济地位”进行调整。我们确实考虑了母亲的种族/民族和教育（后者作为社会经济地位的代表）以及其他特征，以纳入探索自闭症与ART治疗因素和父母不孕症诊断相关性的模型中。这两个因素在单变量分析中均不显著，因此未纳入模型。然而，在回应巴拉德博士等人的“。在此基础上，我们对模型进行了修正，增加了这两个因素。正如预期的那样，ICSI与自闭症诊断相关性的风险估计值在单胎[校正的风险比，aHRR 1.73（1.11-2.69）]和多胎[aHRR 1.51（1.06-2.15）]中均保持统计学显著性。与Barad博士等人的观点相反。尽管社会经济状况良好的患者会在更高程度上使用ICSI，但没有证据表明ICSI的使用与种族/种族（70%的接受ICSI的ART使用者为非西班牙裔白色人，相比之下，70%的未接受ICSI的ART使用者为非西班牙裔白人）或社会经济状况（89%的ICSI使用者为大学毕业生，相比之下，87%的未接受ICSI的ART使用者为大学毕业生）显著相关。在研究期间，ICSI的使用急剧增加，而ART受孕儿童的自闭症诊断却没有类似的增加，这可能是因为自闭症的患病率相对较低。使用数学评估模型来估计个体风险因素的患病率变化对自闭症患病率的影响（Schieve等人，2011年），我们估计，ICSI使用的82%的增加预计将导致孤独症在单胞胎中增加14.4%，因此患病率从0.9%增加到1.03%。同样，我们观察到的ICSI使用增加预计将导致多胞胎中自闭症增加15.5%，这意味着多胞胎中的患病率从1.2%增加到1.39%。自闭症患病率的这种小幅增长很难精确检测。此外，虽然ICSI使用率的增加可能导致自闭症患病率略有增加，但其他因素，如ART分娩中多胞胎的下降，可能会影响相反方向的趋势。正如我们在文章中明确指出的那样，ICSI与自闭症相关的潜在生物学机制尚不清楚，可能与手术本身、选择进行手术的患者的特征或其他因素有关。然而，风险的可能性和缺乏令人信服的获益证据似乎不利于在没有男性因素不孕症诊断的情况下使用ICSI（ASRM，2012）。
Dr Barad et al. express concern that the analysis was not adjusted ‘for race/ethnicity and/or socioeconomic status.’We did consider both maternal race/ethnicity and education (the latter as a proxy for socioeconomic status) among other characteristics for inclusion in models that explored the association of autism with ART treatment factors and parental infertility diagnosis. Neither factor was significant in the univariable analysis and, therefore, was not included in in the models. However, in response to Dr Barad et al.’s suggestion, we refit the models with addition of these two factors. As expected, the risk estimates for the association of ICSI with autism diagnosis remained statistically significant for both singletons [adjusted hazard risk ratio, aHRR 1.73 (1.11–2.69)] and multiples [aHRR 1.51 (1.06–2.15)]. Contrary to Dr Barad et al.’s surmise that socioeconomically advantaged patients would utilize ICSI to a higher degree, there was no evidence that utilization of ICSI was significantly associated with either race/ethnicity (70% of ART users with ICSI were non-Hispanic white, compared with 70% of ART users without ICSI) or socioeconomic status (89% of ICSI users were college graduates, compared with 87% of women who used ART without ICSI). The dramatic increase in use of ICSI during the study period without similar increase in autism diagnosis in ART-conceived children can likely be explained by the relatively low prevalence of autism. Using a mathematical assessment model for estimating the effects of changes in the prevalence of individual risk factors on autism prevalence (Schieve et al., 2011), we estimate that the 82% increase of ICSI use would be expected to result in a 14.4% increase in autism among singletons, and thus a prevalence increase from 0.9 to 1.03%. Likewise, the ICSI use increase we observed would be expected to result in a 15.5% autism increase among multiples, which translates to a prevalence increase from 1.2 to 1.39% among multiples. Such small increases in autism prevalence are difficult to detect with any precision. Moreover, while increasing ICSI utilization may have led to a slight increase in autism prevalence, other factors, such as the decline in multiple births among ART deliveries, might have affected the trend in the opposite direction. As we clearly noted in our article, an underlying biological mechanism through which ICSI could be associated with autism is not known and may be related to the procedure itself, characteristics of the patients selected for the procedure, or other factors. However the possibility of risk and the lack of convincing evidence of benefit do not appear to favor use of ICSI without male-factor infertility diagnosis (ASRM, 2012).