Morales Daniel R, Slattery Jim, Evans Stephen, Kurz Xavier
Pharmacovigilance and Epidemiology Department, European Medicines Agency, 30 Churchill Place, Canary Wharf, London, E14 5EU, UK.
Division of Population Health Sciences, University of Dundee, Dundee, UK.
BMC Med. 2018 Jan 15;16(1):6. doi: 10.1186/s12916-017-0993-3.
Antidepressant exposure during pregnancy has been associated with an increased risk of autism spectrum disorder (ASD) and attention deficit hyperactivity disorder (ADHD) in several observational studies. We performed a systematic review of these studies to highlight the effect that important methodological limitations have on such analyses and to consider approaches to the conduct, reporting and interpretation of future studies.
A review of MEDLINE and EMBASE identified case-control, cohort and sibling studies assessing the risk of ASD and ADHD with antidepressant use during pregnancy. Approaches to confounding adjustment were described. Crude and adjusted effect estimates for comparisons between antidepressant exposure during pregnancy vs. all unexposed women were first meta-analysed using a generic inverse variance method of analysis, followed by effect estimates for alternative pre-selected comparison groups.
A total of 15 studies measuring ASD as an outcome (involving 3,585,686 children and 40,585 cases) and seven studies measuring ADHD as an outcome (involving 2,765,723 patients and 52,313 cases) were identified. Variation in confounding adjustment existed between studies. Updated effect estimates for the association between maternal antidepressant exposure during pregnancy vs. all unexposed women remained statistically significant for ASD (adjusted random-effects risk ratio [RaRR] 1.53, 95% confidence interval [CI] 1.31-1.78). Similar significant associations were observed using pre-pregnancy maternal antidepressant exposure (RaRR 1.48, 95% CI 1.29-1.71) and paternal antidepressant exposure during pregnancy (1.29, 95% CI 1.08-1.53), but analyses restricted to using women with a history of affective disorder (1.18, 95% CI 0.91-1.52) and sibling studies (0.96, 95% CI 0.65-1.42) were not statistically significant. Corresponding associations for risk of ADHD with exposure were: RaRR 1.38, 95% CI 1.13-1.69 (during pregnancy), RaRR 1.38, 95% CI 1.14-1.69 (during pre-pregnancy), RaRR 1.71, 95% CI 1.31-2.23 (paternal exposure), RaRR 0.98, 95% CI 0.77-1.24 (women with a history of affective disorder) and RaRR 0.88, 95% CI 0.70-1.11 (sibling studies).
Existing observational studies measuring the risk of ASD and ADHD with antidepressant exposure are heterogeneous in their design. Classical comparisons between exposed and unexposed women during pregnancy are at high risk of residual confounding. Alternative comparisons and sibling designs may aid the interpretation of causality and their utility requires further evaluation, including understanding potential limitations of undertaking meta-analyses with such data.
多项观察性研究表明，孕期暴露于抗抑郁药与自闭症谱系障碍（ASD）和注意力缺陷多动障碍（ADHD）风险增加有关。我们对这些研究进行了系统综述，以突出重要方法学局限性对此类分析的影响，并探讨未来研究的开展、报告和解释方法。
检索MEDLINE和EMBASE，纳入评估孕期使用抗抑郁药与ASD和ADHD风险的病例对照研究、队列研究和同胞研究。描述了混杂因素调整方法。首先采用通用逆方差分析方法对孕期暴露于抗抑郁药与所有未暴露女性之间比较的粗效应估计值和调整效应估计值进行荟萃分析，随后对其他预先选定的比较组的效应估计值进行分析。
共确定了15项以ASD为结局的研究（涉及3585686名儿童和40585例病例）和7项以ADHD为结局的研究（涉及2765723名患者和52313例病例）。各研究之间在混杂因素调整方面存在差异。孕期母亲暴露于抗抑郁药与所有未暴露女性之间关联的更新效应估计值在ASD方面仍具有统计学意义（调整后的随机效应风险比[RaRR]为1.53，95%置信区间[CI]为1.31 - 1.78）。使用孕前母亲抗抑郁药暴露（RaRR为1.48，95% CI为1.29 - 1.71）和孕期父亲抗抑郁药暴露（1.29，95% CI为1.08 - 1.53）时观察到类似的显著关联，但仅限于有情感障碍病史女性的分析（1.18，95% CI为0.91 - 1.52）和同胞研究（0.96，95% CI为0.65 - 1.42）无统计学意义。ADHD风险与暴露的相应关联为：RaRR为1.38，95% CI为1.13 - 1.69（孕期），RaRR为1.38，95% CI为1.14 - 1.69（孕前），RaRR为1.71，95% CI为1.31 - 2.23（父亲暴露），RaRR为0.98，95% CI为0.77 - 1.24（有情感障碍病史女性）和RaRR为0.88，95% CI为0.70 - 1.11（同胞研究）。
现有测量抗抑郁药暴露与ASD和ADHD风险的观察性研究在设计上存在异质性。孕期暴露组与未暴露组女性之间的经典比较存在残留混杂的高风险。替代比较和同胞设计可能有助于因果关系的解释，其效用需要进一步评估，包括了解对此类数据进行荟萃分析的潜在局限性。
