癲癇懷孕婦女使用抗癲癇藥物與不良產科結果之風險：群組化軌跡模型研究

Abstract

研究背景 癲癇孕婦於孕期使用抗癲癇藥物長久以來是臨床上的難題，需要在控制癲癇與造成胎兒不良結果間取得平衡。過去研究多著重在抗癲癇藥物的致畸胎性，而對於其他不良產科結果的討論不夠全面。此外，多數研究也未善加控制癲癇疾病本身此重大干擾因子，因此可能會影響抗癲癇藥物與不良產科結果相關性的判讀。 研究目的 本研究目的有二，其一為探討孕前穩定使用抗癲癇藥物之癲癇孕婦於懷孕期間不同的抗癲癇藥物使用模式之下，與不良產科結果之相關性。另一為孕前穩定用藥之癲癇孕婦於懷孕期間與產後的抗癲癇藥物使用模式的相關變化。 研究方法 本研究為一回溯性世代研究設計，利用出生通報檔及全民健康保險資料檔，並篩選2004年至2018年孕前穩定使用抗癲癇藥物之癲癇孕婦作為研究對象進行兩階段的分析。第一階段的研究探討抗癲癇藥物的使用與不良產科結果之關聯性。在這一部分的研究中，我們定義研究對象於懷孕前84天至第二孕期（懷孕後196天）為軌跡觀察期，且記錄研究對象於軌跡觀察期間每週是否使用抗癲癇藥物，並以群組化軌跡模式（group-based trajectory model, GBTM）進行分析後，根據不同的用藥模式分組。本研究的不良產科結果包含早產、低出生體重及胎兒小於妊娠年齡，並使用羅吉斯回歸（logistic regression）進行分析，結果以勝算比及95%信賴區間呈現，以探討不同抗癲癇藥物使用模式與不良產科結果的相關性。敏感性分析則帶入不同校正變項的組合，如：產婦年齡、共病症、併用藥物、不良物質使用及妊娠併發症等。第二階段研究則探討上述研究對象產後使用抗癲癇藥物之情形，同樣以群組化軌跡模式方式分出不同的用藥模式。分組完成後，與第一階段研究之軌跡圖比較，並使用桑基圖（Sankey plot）呈現孕前孕期與產後用藥軌跡流動之情形。 研究結果 本研究於第一階段共納入2,175位孕前穩定使用抗癲癇藥物之癲癇孕婦，經群組化軌跡模式分析後，研究對象孕前與孕期的抗癲癇藥物使用模式可分為四組，分別為frequent and continuous users (64.9%)、frequent but discontinuous users (7.1%)、intermittent users (19.7%)、intermittent and discontinuous users (8.3%)。 研究對象孕期的抗癲癇藥物使用模式與不良產科結果皆無顯著相關性，以frequent and continuous users作為對照組時，早產事件經校正後的勝算比依frequent but discontinuous、intermittent、intermittent and discontinuous users的組別順序分別為：0.83 [95% CI: 0.47-1.48]、0.71 [0.47-1.05]、0.88 [0.52-1.49]。低出生體重校正後的勝算比依組序則為：0.89 [0.51-1.56]、0.70 [0.47-1.04]、1.34 [0.84-2.12]。最後，胎兒小於出生體重之校正後的勝算比依組序為：0.95 [0.58-1.57]、0.92 [0.66-1.28]、0.98 [0.62-1.56]。而各不良事件的敏感性分析結果與主分析結果一致。 本研究於第二階段共納入2,171位孕前穩定使用抗癲癇藥物之癲癇孕婦，經群組化軌跡模式分析後，研究對象產後的抗癲癇藥物使用模式可分為三組，分別為frequent users (62.3%)、intermittent users (25.9%)、non users (11.8%)。大多數人於產後會繼續維持與孕期同樣的用藥模式。如孕前及孕期組別為frequent and continuous 或 intermittent者，產後大部分仍是frequent或intermittent users，而孕期停藥的frequent but discontinuous與intermittent and discontinuous users，產後大多未再使用抗癲癇藥物。 研究結論 於本研究中，台灣地區癲癇孕婦孕前與孕期使用抗癲癇藥物的模式可分為四種，不同用藥模式的孕婦之間並未觀察到不良產科結果有顯著差異，因此在醫療人員仔細評估癲癇孕婦用藥的利弊後，可積極介入治療。而癲癇孕婦產後的用藥模式則可分為三種穩定的狀態，本研究發現許多孕前穩定使用藥物但於孕期停藥者，其產後也未恢復藥物使用以控制癲癇。此狀況之原因與後續是否造成不良結果尚待日後研究提供更多證據。

Background Use of antiepileptic drugs (AEDs) during pregnancy for women with epilepsy is a challenge in clinical setting. For these women, a balance between seizure control and adverse obstetric outcomes must be addressed by clinicians. Among all adverse obstetric outcomes associated with AEDs, teratogenicity may be the mostly well-known. However, studies on other adverse obstetric outcomes, such as preterm birth, are very limited. In addition, most existing studies are limited to confounding by indication as they did not consider the underlying epilepsy when investigating the association between AEDs and adverse obstetric outcomes. However, epilepsy itself can be linked to adverse obstetric outcomes. Objectives This nationwide study aims 1) to examine the association between different exposure patterns of AEDs during pregnancy and the risk of adverse obstetric outcomes; and 2) to evaluate the changes of AEDs uses during pregnancy and after delivery among pre-pregnancy chronic AEDs users among women with epilepsy. Method This is a retrospective cohort study retrieving data from the Birth Certificate Application and National Health Insurance Research Database (NHIRD) in Taiwan from 2004 to 2018. Women with epilepsy who were pre-pregnancy chronic AED users were identified as our study cohort. In the first part of the study, we assessed weekly consumption of AEDs among pregnant women with epilepsy and used group-based trajectory modeling (GBTM) to identify distinct groups of AEDs use from 3 months before conception of pregnancy to the second trimester. Adverse obstetric outcomes of interest were preterm birth, low birth weight (LBW), and small for gestational age (SGA). Logistic regressions were used to examine the association between distinct trajectories of AEDs uses and adverse obstetric outcomes. Sensitivity analyses were conducted with different covariates adjusted in the regression models, such as maternal age, comorbidities, comedications, hazardous substance use, and pregnant complications. In the second part, we also conducted GBTM to describe the AEDs uses after delivery among the aforementioned study cohort. Sankey plot was used to visualize the changes of trajectories of AEDs uses during pregnancy (first part) and after delivery (second part). Results Of the 2,175 eligible pregnant women, we identified four trajectories of AEDs uses from 3 months before conception to the second trimester: frequent and continuous users (64.9%), frequent but discontinuous users (7.1%), intermittent users (19.7%), intermittent and discontinuous users (8.3%). The associations between different trajectories of AEDs uses and adverse obstetric outcomes were not statistically significant when using frequent and continuous users as the reference group. For preterm, the adjusted odds ratios (ORs) of frequent but discontinuous users, intermittent users, intermittent and discontinuous users were 0.83 [95% CI: 0.47-1.48], 0.71 [0.47-1.05], 0.88 [0.52-1.49], respectively. For LBW, the adjusted ORs were 0.89 [0.51-1.56], 0.70 [0.47-1.04], 1.34 [0.84-2.12], respectively. For SGA, the adjusted ORs were 0.95 [0.58-1.57], 0.92 [0.66-1.28], 0.98 [0.62-1.56]. Results of sensitivity analyses were similar with the main analyses. In the second part of the study, we identified three trajectories of AEDs uses during one year after delivery: frequent users (62.3%), intermittent users (25.9%), and non-users (11.8%). Most of the women with epilepsy in our study maintained the same AEDs utilization pattern during pregnancy and after delivery. For example, most of the frequent and continuous users or intermittent users during pregnancy were frequent users or intermittent users after delivery. Also, most of the women who discontinued AEDs during pregnancy never use AEDs again after delivery. Conclusion We identified four distinct trajectories of AEDs use among pregnant women in Taiwan, and no significant difference between the association between distinct trajectories of AEDs use and risk of adverse obstetric outcomes was found in this study. Therefore, women with epilepsy could receive AEDs after health professionals evaluate the risks and benefits. Three trajectories of AEDs use after delivery among women with epilepsy were also identified. Our study revealed that for those who discontinue AEDs during pregnancy, they did not add back AEDs after delivery even they were chronic AEDs users before pregnancy. The reason and the consequence of this situation should be discussed in future studies.