使用抗癲癇藥物引發青光眼的風險

Abstract

目的:計算使用抗癲癇藥物的患者罹患青光眼的風險並探討兩者之間的相關性 資料來源: 台灣全民健保研究資料庫 材料與方法: 使用的資料是來自於全民健保研究資料庫自2006年1月1日起至2007年12月31日止所有病人門診和住院的就醫資料。起初，先揀選擁有記錄ICD-9-CM為365並同時使用青光眼藥物的病患，並將其定義為青光眼患者，而第一次擁有青光眼的診斷且合併青光眼用藥的就診日期定義為index date。排除前一年即是青光眼病患者、排除性別不明者並排除在index date之前一個月內曾接受過白內障手術、玻璃體切除術以及玻璃體內注射的病患，最後得到我們所需要的研究個案。計算所有研究個案的平均年紀和性別比例的分布情形。在case-crossover的研究架構下，我們分別將case period和control period定義為index date前1-30天和index date前91-120天，並且校正其他常用的抗癲癇藥物以及與青光眼發生有關聯性的藥物。我們依照不同的治病機轉將抗癲癇藥物分為三大類分別計算這三大類藥物引發青光眼的風險，接著再計算各個抗癲癇藥物的風險做進一步的分析。 結果: 在case period和control period期間，使用topiramate的discordant pairs分別為31和26。粗勝算比為1.19(95% C.I.為0.71-2.01)，校正了其他抗癲癇藥物和同時期的其他用藥之後，其校正後的勝算比則為1.21( 95% CI 0.72-2.05)。而將所有抗癲癇藥物分為三類分別計算其校正後的勝算比，可得到鈉離子阻斷劑類的藥物勝算比為1.15 (95% C.I. 1.01-1.31)，GABAergic藥物和GABAA 受器拮抗劑則為1.18 (95%C.I. 1.07-1.29)。至於針對每一種個別藥物計算之後，則只有oxcarbazepine和clonazepam的風險值有升高並達統計顯著意義。另外β受器阻斷劑則呈現些許的保護作用(校正後的勝算比 0.97, 95% C.I. 0.92-1.01)。 結論: 雖然topiramate引發青光眼的校正後勝算比呈現未達統計顯著意義的升高，但這依然是一個雖然少見卻可能造成永久性失明的嚴重藥物副作用，也已經被美國藥物與食物管理局列為該藥物的副作用之一，又好發於中年人，所以依然值得被關切。此外，我們的研究結果也發現oxcarbazepine這個藥物也造成罹患青光眼的風險值有升高的情形，據我們推測認為是oxcarbazepine所引起的青光眼類型與topiramate所引發的不同才未引發大眾的關注。從case-crossover的研究設計中可計算出勝算比但無法得知發生率以及危險因子，若想解決上述所提到的兩的問題則必須考慮做世代研究來加以釐清。

Purpose: To estimate the risk of glaucoma attack for patients under the treatment of anticonvulsants Data Source: National Health Insurance Research Database (NHIRD) Material and Methods: To identify the glaucoma cases newly diagnosed during the January 1, 2006 to December 31, 2007 from the NHIRD. At first, we identified the glaucoma patients were those who were coded with ICD-9-CM of 365 combined with antiglaucomatous medication use, and the first recorded date was defined as index date. Then, the glaucoma patients were diagnosed during the year before the index date, the patients with unknown gender and the secondary glaucoma induced by the intraocular surgery were all excluded. After excluding those who were not meet the inclusion criteria, the residual patients were defined as cases. The proportion of gender and the mean age were calculated. We defined the time window was 30 days and the case period and the control period as 1 to 30 days and 91-120 days before the index date respectively. The concomitant drug use was analyzed and those could affect the incidence of glaucoma would be further adjusted. At first, all of selected anticonvulsants were classified to three categories based on the mechanism of antiepileptic effects and to calculate the crude odds ratio and adjusted odds ratio. Further, the risk of every anticonvulsant was estimated. Results: During the case period and the control period, the numbers of discordant pairs were 31 and 26 respectively. The crude odds ratio was 1.19 (0.71-2.01). The adjusted odds ratio was 1.21 (0.72-2.05) after adjusting the other two categories of anticonvulsants and the concomitant drugs. The adjusted odds ratio of sodium channel blocker and GABAergic drugs or GABAA receptor agonist were 1.15 (1.01-1.31) and 1.18 (1.07-1.29) respectively. As for the adjusted odds ratio of the every anticonvulsant, the patients with oxcarbazepine use had statistically significant increased risk of glaucoma attack. The adjusted odds ratio of β-blocker showed mild protective effect with borderline statistical significance (adjusted odds ratio 0.97, 95% C.I. 0.92-1.01). Conclusions: Although the adjusted odds ratio of topiramate-induced glaucoma showed mildly increased without statistical significance, it still was a rare but severe adverse drug effect and should be noticed and avoided. Besides, the study results also represented that the anticonvulsants use may increase the risk of glaucoma attack with unknown mechanism. We thought the type of glaucoma induced by oxcarbazepine use may be different from that induced by topiramate. Therefore, it was not noticed widespread until now. The case-crossover study design was an appropriate method to estimate the risk for our study, but it could not give us more information about the high risk group and the incidence in Taiwan. If we want to know that, cohort study would be a better study design.