孕期鎮靜安眠藥品暴露與胎兒不良結果之相關性

Abstract

研究背景: 評估懷孕婦女使用鎮靜安眠藥品(benzodiazepines/z-hypnotics)的效益與風險平衡(benefit and risk balance)，長久以來都是臨床醫療人員與衛生主管機關的巨大挑戰。儘管此議題相當重要，過去文獻對於此領域的探討並不夠全面，不僅孕期的使用率資料相當有限且研究樣本數少，對於干擾因子的控制也不夠周詳，因此可能導致研究誤差產生。另外，多數研究高度集中於西方國家，極度缺乏亞洲族群的相關資料。 研究目的: 本研究擬進行兩個主題之研究，分別為 (1) 探討2004-2018年間懷孕婦女於孕期使用鎮靜安眠藥品之變化趨勢和使用族群特性，以及分析懷孕期間之治療模式(treatment pattern)。 (2) 評估孕期20週內之鎮靜安眠藥品暴露與胎兒不良結果發生之相關性，包含死產(stillbirth)、早產(preterm birth)和胎兒小於妊娠年齡(small gestational age)。 研究方法: 本研究為一回溯性世代研究，利用三種資料檔案，包含出生通報檔、全民健康保險資料檔以及婦幼主題式資料庫，並篩選2004年至2018年之單胞胎生產懷孕婦女以進行兩個主題的研究分析。主題一的研究中首先探討孕期鎮靜安眠藥品使用之盛行率，並依照藥品使用類別分成：總體鎮靜安眠藥品使用、BZDs類藥品使用、z-hypnotics類藥品使用、常見的五種鎮靜安眠藥品使用以及併用抗憂鬱藥品或鴉片類止痛藥物之使用盛行率。其次，依序利用單變項迴歸分析及多變項迴歸分析，以了解孕期鎮靜安眠藥品使用之相關風險因子。最後，本研究將採取間斷區間時間序列設計(interrupted time series design)，藉以評估不同孕期以及生產對於鎮靜安眠藥品使用盛行率(level)以及趨勢(trend)的影響。而主題二的研究中，根據懷孕婦女是否於懷孕開始日至懷孕開始日後第140天具有至少一次的鎮靜安眠藥品領藥紀錄分為暴露組與非暴露組，以探討胎兒死產、早產以及胎兒小於妊娠年齡之發生風險。暴露組與非暴露組將進行propensity-score fine stratification weighting (PS-FSW)，並進一步利用適應症干擾控制研究設計、手足對照研究設計(sibling control analyses)以及父親負向控制研究設計(paternal negative control analyses) 以降低干擾因子影響。本研究利用羅吉斯迴歸(logistic regression model)計算鎮靜安眠藥品暴露與不良結果發生之相關性。結果以勝算比(odds ratio)及95%信賴區間(confidence interval)呈現，分析採取雙尾檢定，p-value<0.05視為具有統計上之顯著。 研究結果: 本研究於主題一共納入2,883,162位單胞胎生產紀錄，並根據孕期是否暴露於鎮靜安眠藥品，分成100,737位暴露者，和2,782,425位非暴露者。孕期鎮靜安眠藥品之總體盛行率從2004年的3.7%維持穩定趨勢至2018年的3.3%，十五年間總盛行率為3.5%，而併用抗憂鬱藥品或鴉片類止痛藥物之情形，盛行率隨著時間大幅上升。孕期鎮靜安眠藥品使用潛在因子分析則發現孕婦年齡較小、不健康的生活行為、較高的孕期疾病負擔、具有睡眠障礙或精神相關疾病以及孕前較頻繁的醫療利用行為，與孕期使用鎮靜安眠藥品具較高的關聯性。另外，探討不同孕期介入影響則顯示，第一孕期鎮靜安眠藥品的盛行率迅速下降，盛行率變化為-0.59% (95% CI: -0.64% to -0.55%)，趨勢變化為-0.05% (95% CI: -0.07% to -0.03%)。然而，生產後的使用率顯著上升，盛行率變化為0.12% (95% CI: 0.07% to 0.16%)，趨勢變化為0.04% (95% CI: 0.03% to 0.06%)。 而在本研究的主題二中共收納2,882,292位單胞胎生產紀錄，其中暴露組共75,679人，占研究族群的2.6%。於死產結果發生率方面，暴露組之發生率為每百人1.25件，未暴露組則為每百人0.88件。而於早產結果方面，暴露組發生率為每百人9.95件，未暴露組為每百人7.04件。最後，針對胎兒小於妊娠年齡發生的部分，暴露組的發生率為每百人11.24件，非暴露組為每百人9.61件。 不良結果發生風險方面，相比於未暴露組，在考慮母親基本特性及共病症後，暴露組將輕微增加胎兒發生死產(wOR 1.19, 95% CI 1.10-1.28)、早產(wOR 1.19, 95% CI 1.16-1.23)及胎兒小於妊娠年齡(wOR 1.16, 95% CI 1.13-1.19)之風險。然而，進一步控制適應症干擾後，則未發現孕期藥品暴露與胎兒發生死產具顯著關聯性。而在利用手足對照分析校正家庭間潛在干擾因子和基因遺傳之影響後，則發現於孕期暴露鎮靜安眠藥品的孕婦，僅將輕微增加新生兒發生胎兒小於妊娠年齡之風險(aOR 1.13, 95% CI 1.06-1.20)，然而並無顯著增加死產(aOR 1.09, 95% CI 0.95-1.26)或早產發生的風險(aOR 0.95, 95% CI 0.89-1.01)；父親負向控制研究之結果亦進一步驗證手足對照分析之結果可降低家庭與遺傳相關干擾因子產生的偏誤。 研究結論: 儘管孕期鎮靜安眠藥品之使用於十五年追蹤期間維持穩定，然而併用抗憂鬱藥品或鴉片類止痛藥物之情形日益俱增，衛生主管機關須適時進行相關政策的制定及宣導。另外，針對具有用藥風險因子的懷孕婦女，臨床醫療人員須進一步關注該族群以判定治療決策之介入時機。最後，對於懷孕初期停用鎮靜安眠藥品之懷孕婦女，臨床醫師於照護上須密切監測因停藥導致的戒斷症狀和精神相關疾病復發之情形，也應特別注意產後用藥者的產後憂鬱，或者焦慮對於健康造成的不良結果，以及哺乳過程藥物透過乳汁對於嬰兒造成之影響。 而在胎兒不良結果發生風險方面，本研究發現於孕期早期暴露鎮靜安眠藥品，僅輕微增加新生兒發生胎兒小於妊娠年齡之風險，並無顯著增加死產或早產之發生。雖然上述不良結果不具有明顯的臨床相關性，然而鎮靜安眠藥品僅適合於全面考量用藥的效益與風險平衡後，才得以適時介入懷孕婦女之睡眠症狀以及精神相關疾病，以對於個體做出最適當的治療。

Background: Evaluation of the benefits and risks of pregnant women using benzodiazepines (BZDs) and z-hypnotics has long been a great challenge for clinicians and health authorities. Although it is essential to investigate the drug safety for pregnant women, little is known about the adverse neonatal outcomes associated with maternal use of BZDs/Z-hypnotics. Existing studies were limited to a small study population and unmeasured confounders, which may lead to research bias. In addition, studies in the Asian population are scarce because most studies were conducted in Western countries. Objectives: The study aims to assess the use of BZDs/Z-hypnotics during pregnancy and the associated risk of adverse neonatal outcomes. The study is divided into the following two subsections: (i) to examine the prevalence trends and utilization patterns of BZDs and z-hypnotics among pregnant women in Taiwan; (ii) to evaluate the association between prenatal BZDs and z-hypnotics exposure and the risk of stillbirth, preterm birth, and small for gestational age (SGA). Methods: This is a population-based retrospective cohort study using data from three databases, the National Birth Certificate Application (BCA) database, the National Health Insurance Research Database (NHIRD), and the Maternal and Child Health Database (MCHD). Singleton pregnancies with a gestational age of at least 20 weeks between 2004 and 2018 were included. In the first part of the study, the prevalence trends of BZDs/z-hypnotics used during pregnancy were estimated annually, and the proportions of co-medication with antidepressants and opioids were also evaluated across the study period. Multivariable logistic regression models were used to compare the risk factors associated with BZD/z-hypnotics users during pregnancy. The utilization patterns of BZDs/z-hypnotics (I) before pregnancy; (II) after the first trimester; (III) after the second trimester; (IV) after the third trimester; and (V) after birth were assessed using interrupted time series analysis (ITSA). Segmented regression models were further used to estimate the level and trend changes of BZDs/z-hypnotics use across different stages of pregnancy. In the second part of the study, a nationwide cohort study was conducted to evaluate the risk of stillbirth, preterm birth, and SGA following early pregnancy exposure to BZDs/z-hypnotics. The main analyses controlled for maternal covariates and confounding by indication. Pre-specified sensitivity analyses such as sibling control analyses and paternal negative control analyses were used to test the robustness of our analyses. Logistic regression models using propensity score fine stratification weighting were applied to assess the risk of adverse neonatal outcomes. The odds ratio (OR) and 95% confidential interval (95% CI) were reported, and a two-tailed p-value <0.05 was considered statistically significant. Results: In the first part of the study, BZDs/z-hypnotics use among 2,883,162 pregnant women was 3.5% and remained stable from 2004 to 2018 (3.7% to 3.3%). The concurrent use among BZDs/z-hypnotics users increased 3-folds for antidepressants (8.8% to 23.6%) and 2-folds for opioids (0.9% to 2.2%) across the study period. Pregnant women with lower maternal age, unhealthy lifestyle behavior, higher obstetric comorbidity index, psychiatric disorder, and higher healthcare utilization were associated with a higher probability of BZDs/z-hypnotics use during pregnancy. In addition, the ITSA revealed a sudden decrease of BZDs/z-hypnotics uses at 2 weeks after conception (level change -0.59%, 95% CI -0.64% to -0.55%; trend change -0.05%, 95% CI -0.07% to -0.03%). In contrast, the BZDs/z-hypnotics use increased remarkably after the delivery of these women (level change 0.12%, 95% CI 0.07% to 0.16%; trend change 0.04%, 95% CI 0.03% to 0.06%). In the second part of the study, the cohort included 2,882,282 singleton pregnancies, and 75,679 (2.6%) were dispensed one or more BZDs/z-hypnotics during the early pregnancy. The incidence rate of stillbirth was 1.25 per 100 pregnancies among exposed women versus 0.88 per 100 pregnancies among unexposed women. In terms of preterm birth, the incidence rate of the exposed women was 9.95 cases per 100 pregnancies, and the unexposed women were 7.04 cases per 100 pregnancies. For SGA, the incidence rate was 11.24 and 9.61 per 100 pregnancies, respectively. Early pregnancy exposure to BZDs/z-hypnotics was associated with adverse neonatal outcomes as compared with unexposed offsprings. The PS-weighted ORs (wOR) was 1.19 (95% CI 1.10-1.28) for stillbirth, 1.19 (95% CI 1.16-1.23) for preterm birth, and 1.16 (95% CI 1.13-1.19) for SGA. However, in models that further control for confounding by indication, the risk of stillbirth associated with BZDs/z-hypnotics exposure was attenuated. Moreover, in models with sibling controls that account for familial confounding and genetic factors, early BZDs/z-hypnotics exposure was associated with a small increase in the risk of SGA (aOR 1.13, 95% CI 1.06-1.20) but not with stillbirth (aOR 1.01, 95% CI, 0.81-1.25) and preterm birth (aOR 0.95, 95% CI 0.89-1.01). Results from paternal negative control analyses were consistent with findings from the sibling control analyses. Conclusions: The prevalence of BZDs/z-hypnotics exposure during pregnancy remained stable during 2004 and 2018, but the longitudinal trends of co-medication with antidepressants and opioids continued to increase, which warrants clinical or policy implementations. Furthermore, women who discontinue BZDs/z-hypnotics after conception should be monitored closely for withdrawal syndromes and mental illness relapse, and those who use BZDs/z-hypnotics after the delivery should be monitored regularly for postpartum depression or anxiety. In terms of adverse neonatal outcomes, our findings suggest that BZDs/z-hypnotics used in early pregnancy are not associated with a substantial increase in risk for stillbirth and preterm birth. However, clinicians should be aware of the potential increased risk of SGA.