利用跨平台資料庫評估水痘疫苗之潛在風險

Evaluating the safety of varicella vaccine by using information from different claims data

劉家鴻(Chia-Hung Liu)

指導教授 ： 陳建煒

共同指導教授 ： 季瑋珠(Wei-Chu Chie)

國立臺灣大學/公共衛生學院/流行病學與預防醫學研究所/博士(2020年)

https://doi.org/10.6342/NTU202001073

摘要

背景
疫苗上市後的安全訊號(safety signals)偵測及精練(detection and refinement)面臨缺乏主動偵測及研究方法限制等問題。國內自2004年開始全面接種水痘疫苗，但目前國內外對水痘疫苗安全性研究並不多。本研究串連國內高涵蓋率的全民健保研究資料庫 (National Health Insurance Research Database)醫療照護記錄及全國預防接種資訊管理系統(National Immunization Information System)，進行水痘疫苗上市後之主動偵測的資料庫探勘(data mining)及安全訊號之檢定。

研究方法
透過串連台灣全民健保研究資料庫的醫療照護記錄及全國預防接種資訊管理系統的疫苗資料，建立2004-2014年水痘疫苗接種世代。研究對象納入條件為12至35個月之幼童第一次接受水痘疫苗族群，排除同一天接受其他疫苗者。
資料庫探勘法以self-controlled tree-temporal scan statistic方式進行，疫苗注射後1-56天內急診及住院診斷納入潛在不良事件，事件發生前183天內無相似診斷。
安全訊號以Self-controlled risk interval design針對疫苗注射後預先設定的住院不良事件進行檢定(pneumonia, idiopathic thrombocytopenic purpura [ITP], menigitidis, encephalitits及ischemic stroke)，定義新事件為發生前14天無相同診斷。危險期間為疫苗接種後1-42天，對照期間為43-84天，同時校正二個時間變量：年齡及季節。以Conditional Poisson regression比較危險期間及對照期間內不良事件之發生率(incidence rate)。

結果
2004年1月至2014年9月，在排除872,600位同一天接受其他疫苗後，共有1,194,189位僅接受單一水痘疫苗接種之幼童進入主要分析。
資料庫探勘發現五個statistical signals：fever in days 1-6 (relative risk (RR) 1.24, p<0.001)、gastritis and duodenitis in days 1-2 (RR 1.9, p<0.001)、acute respiratory infections in days 1-6 (RR 1.43, p<0.001)、varicella infection (RR 4.95, p<0.001)及nausea and vomiting in days 1-3 (RR 1.4, p=0.042)，安全訊號皆發生在疫苗注射後10天內。
疫苗注射後1-42天內，未觀察到特定不良事件有增加危險，但觀察到immunosuppressive幼童有pneumonia上升的風險(adjusted incidence rate ratio [aIRR] 1.29; 95% confidence interval [CI], 1.02-1.62)。另外，36-42天有pneumonia 發生上升的危險 (aIRR 1.10; 95% CI, 1.02-1.18)。

結論
利用台灣全民健保資料庫的醫療照護記錄及全國預防接種資訊管理系統資料，有助於疫苗上市後安全性研究之進行。我們並未偵測到水痘疫苗注射後有未知的安全訊號。對於已知可能潛在不良反應，觀察到疫苗注射第六週有肺炎上升的趨勢。

關鍵字

水痘疫苗 ； 疫苗安全性評估 ； 資料庫探勘 ； 自我控制危險期間分析法 ； 藥物流行病學

並列摘要

Background
Even though the current proof of varicella vaccine safety is made of from the results of premarketing clinical trials and rare reported severe adverse events, it still needs the comprehensive epidemiologic evidence of post-marketing study to support the evidence of safety of universal varicella vaccination. With the linkage of the Taiwan National Health Insurance data and the National Immunization Information System, we performed active safety surveillance after varicella immunization and evaluated the association of pre-specified adverse events and varicella vaccines.

Method
The study population was 12- to 35-month children who received varicella vaccine during 2004 January through 2014 September.
In this post-vaccination surveillance study with data mining method, we collected all incident ICD-9-CM diagnoses in emergency or inpatient department occurring 1–56 days after varicella vaccination. We used TreeScan in a conditioned self-controlled tree-temporal method to explore clustering events in groups with potential temporal risk intervals. The comparison interval consisted of days outside the risk interval.
The self-controlled risk interval design compared the incidence of pre-specified outcomes of interest during a risk interval of 1-42 days post-vaccination and a control interval of days 43-84. The outcomes of interest were defined as admitted pneumonia, idiopathic thrombocytopenic purpura, meningitis, encephalitis, and ischemic stroke, as well as fracture as a negative control. Conditional Poisson regression was used to assess the incidence rate ratio (IRR) with adjustments for age and seasonal effects.

Result
Of the 2,003,861 children receiving varicella vaccines, 1,194,189 (59.6%) children who did not receive other vaccines at the same time. Among them, there were 130,617 events occurring in 56 days after vaccination. We identified 9 alerting nodes with clustering features using TreeScan. The alerting nodes were then categorized into four safety signals: fever at day 1-6 (ICD-9 780.6, relative risk (RR) 1.24, p<0.001), gastritis and duodenitis at day 1-2 (ICD-9 535.00, RR 1.79, p<0.001), acute upper respiratory infections at day 1-5 (ICD-9 465.9, RR 1.62, p=0.006), and varicella infection without mention of complication at day 1-9 (ICD- 052.9, RR 4.96, p<0.001). All these signals and their risk periods have been identified in previous clinical trials and surveillance studies.
In self-controlled risk interval study, there was no increase in the risk for ITP (aIRR 1.00; 95% CI, 0.76-1.33), meningitis (aIRR 1.21; 95% CI, 0.49-2.95), encephalitis (aIRR 1.00; 95% CI, 0.62-1.60), or ischemic stroke (aIRR 1.24; 95% CI, 0.31-4.95). A clustering feature with pneumonia was detected during days 36-42 post-vaccination (aIRR 1.10; 95% CI, 1.02 -1.18) and in children with immunosuppression during days 1-42 post-vaccination (aIRR 1.29; 95%CI, 1.02-1.62).

Conclusion
We did not detect unexpected acute medical events in cluster after mass varicella vaccination in Taiwan. A slightly increase in the risk of incidental pneumonia associated with varicella vaccine was observed in the 6th week after immunization.

並列關鍵字

Varicella vaccine ； data mining ； self-control risk interval design ； self-controlled tree-temporal scan statistic

參考文獻

1. Shimabukuro TT, Nguyen M, Martin D, DeStefano F. Safety monitoring in the Vaccine Adverse Event Reporting System (VAERS). Vaccine. 2015;33(36):4398-405.

2. Intussusception among recipients of rotavirus vaccine--United States, 1998-1999. MMWR Morb Mortal Wkly Rep. 1999;48(27):577-81.

3. Iskander JK, Miller ER, Chen RT. The role of the Vaccine Adverse Event Reporting system (VAERS) in monitoring vaccine safety. Pediatr Ann. 2004;33(9):599-606.

4. Baggs J, Gee J, Lewis E, Fowler G, Benson P, Lieu T, et al. The Vaccine Safety Datalink: a model for monitoring immunization safety. Pediatrics. 2011;127 Suppl 1:S45-53.

5. Yu O, Bohlke K, Hanson CA, Delaney K, Rees TG, Zavitkovsky A, et al. Hepatitis B vaccine and risk of autoimmune thyroid disease: a Vaccine Safety Datalink study. Pharmacoepidemiol Drug Saf. 2007;16(7):736-45.