小兒氣喘在次級資料中之嚴重度模型建立與驗證

Development and Validation of Severity Model for Childhood Asthma in Claims Data

陸翊豪(Yi-Hao Lu)

指導教授 ： 鍾國彪

國立臺灣大學/公共衛生學院/健康政策與管理研究所/碩士(2019年)

https://doi.org/10.6342/NTU201900381

摘要

研究背景與目的：氣喘是國人常見的慢性呼吸道疾病，氣喘死亡率與氣喘嚴重度之間存在著相關性，必須藉由氣喘嚴重度的分級評估來採取不同的治療策略，因此氣喘嚴重度在臨床上以及氣喘品質照護上扮演很重要的角色。根據台灣氣喘諮詢協會出版的氣喘診療指引指出，兒童與青少年為氣喘的主要族群，而成人與兒童也應採取不同的評估方法，因此需要發展適合小兒氣喘的嚴重度分級模式。目前，嚴重度的分級尚未有明確的共識，在過去也有許多的準則建立試圖來填補次級資料嚴重度缺口，但沒有一個統一且最好的準則，必須整合許多準則及資料，甚至有嚴重度發展方法上使用了多個準則進行次級資料的相關嚴重度分級模式，同時，也必須針對每個國家的情況進行準則修正，因此本研究欲發展屬於台灣次級資料的小兒氣喘嚴重度分級模式。準則上的嚴重度分類十分複雜以及多樣化，對於其對於其準則的效度驗證也是必要的，過去針對次級資料所做的嚴重度分級模式，其建立後的效度驗證卻是十分缺乏的，因此在嚴重度分級模式建立後的驗證分析是值得探討以及研究，才能對於發展的次級資料嚴重度分級模式具有一定的效果以及適配度。總結上述，本研究的目的在於探討並建立次級資料小兒氣喘嚴重度分級模式，以及對於小兒氣喘嚴重嚴重度分級模式進行驗證。

材料與方法：本研究採用全民健康保險研究資料庫，確立2011年之小兒氣喘新發個案，根據新發個案的病人其氣喘就醫紀錄當中的氣喘相關用藥紀錄進行分類，建立分類後的用藥進行每次就醫醫師所開立的用藥組合，將各種用藥組合進行三種嚴重度用藥組合分類，再根據每個病人三類嚴重度用藥組合的比例進行病人的嚴重度分級建立，利用各種閾值方式判定用藥比例的標準篩選，最後分出各級嚴重度。之後進行分級模式的相關統計分析，包括存活分析使用Kaplan-Meier曲線、羅吉斯迴歸以及Cox Proportional Hazard Model探討嚴重度及急性發作之間的風險關係，以及分級模式建立後的分析驗證，測量嚴重度分級模式的適配性，最後發展出適合小兒氣喘的次級資料嚴重度分級模式。
研究結果：在嚴重度分級模式中，觀察各閾值存活曲線以及專家的討論後，認為採用閾值50%作為後續嚴重度用藥組合比例的閾值較為適當，後續的嚴重度分析上發現嚴重度與急性發作在存活分析曲線各組間存在顯著差異，嚴重度越大的小兒氣喘病人，其急性發作發生的風險越高。在經過合併級別成三級及兩級之後，嚴重度與急性發作之間的存活分析曲線也存在顯著差異，嚴重度越大的氣喘病人，其急性發作的發生風險也越高。而在分級模式的驗證上，嚴重度與急性發作在存活分析曲線上各級趨近顯著，在經過合併級別成三級及兩級之後，嚴重度與急性發作在存活分析曲線上各級達到顯著差異。

結論與建議：本研究探討氣喘嚴重度的分級模式建立，在無法獲得初級臨床資料下，發展出適合小兒氣喘的次級資料嚴重度分級模式，得到良好的研究結果。在利用不同年度進行分級模式的模型驗證，利用相同的模型套用至不同年度的次級資料也具有相同的適用程度，此方法的建立後可在缺乏臨床資料下也能建立屬於小兒氣喘的嚴重度分級模式。對未來的建議包含可將小兒氣喘嚴重度分級模式用於後續的研究應用上，例如氣喘給付改善方案等論質計酬的相關研究。

關鍵字

氣喘 ； 小兒氣喘 ； 氣喘嚴重度 ； 分級模式 ； 驗證 ； 次級資料 ； 閾值

並列摘要

Background and purpose: Asthma is a common chronic respiratory disease. There is a correlation between asthma mortality and asthma severity. Different treatment strategies must be adopted by grading assessment of asthma severity. Therefore, asthma severity plays a very important role in the care of asthma and quality. According to the guidelines for asthma diagnosis and treatment published by Taiwan Asthma Council, children and adolescents are the main ethnic groups of asthma. Adults and children should also adopt different assessment methods. Therefore, it is necessary to develop a severity classification model suitable for children with asthma. At present, there is no clear consensus on the grading of severity. In the past, there were many guidelines attempt to establish and fill the gap in the severity of secondary data. However, there is no uniform and best criterion. Many criteria and materials must be integrated. Also, some development methods use many criteria to develop a severity model of secondary data. At the same time, it must also be revised for each country. Therefore, this study aims to develop a pediatric asthma severity classification model that belongs to Taiwan's secondary data. The classification of severity on the criteria is very complicated and diversified, and it is also necessary for its validation of the criteria. In the past, the severity classification model for secondary data is lack in validation after establishment. Therefore, the validation analysis after the establishment of the severity classification is worthy of discussion and research, in order to have a certain effect and adaptability to the development of the secondary data severity classification model. To sum up, the purpose of this study is to explore and establish a secondary data classification model for asthma in children, and to verify the classification of severe asthma severity in children.

Materials and Methods: This study used National Health Insurance Research Database to establish a new case of asthma in children in 2011. According to the new cases, the patients were classified according to the asthma-related medication records, and the classified medications divided into three types of drug combinations group, and the patient's severity classification is established according to the ratio of the three types of severity medication groups of each patient, and the threshold ratio develops to classify three severity level. Subsequent statistical analysis of the grading model, including survival analysis using Kaplan-Meier curves, Logistic regression and Cox Proportional Hazard Model to explore the risk relationship between severity and acute exacerbation, and the validation after grading model establishment, adaptability of the severity model, and finally developed a severity model suitable for children with asthma in secondary data.

Results: After observing the threshold survival curves and the expert's discussion, it is considered to use the threshold of 50% as the threshold for standard of the drug group ratio, and the severity analysis found association between severity and acute exacerbation. The results found that the greater the severity of pediatric asthma patients, the higher the risk of acute exacerbation. There is also a significant difference in the survival analysis curve between severity and acute exacerbation after the combined level of three levels. In the validation of the model, the severity and acute exacerbation are significantly at all levels on the survival analysis curve. After the merging level becomes three levels, the severity and acute exacerbation are significant at all levels on the survival analysis curve.

Conclusion and suggestions: This study explored the establishment of a classification model for asthma severity. Under the condition that primary clinical data could not be obtained, a secondary data severity classification model suitable for children with asthma was developed, and good results were obtained. In the model validation using the different years of the model, the same model can be applied to the secondary data of different years would have the same applicability. After the establishment of this method, the severity of pediatric asthma can be established in the absence of clinical data. Suggestions for the research include the use of the pediatric asthma severity model for further research applications, such as the policy of Pay for Performance and the other related applications.

並列關鍵字

Asthma ； Pediatric asthma ； Asthma severity ； Severity model ； Validation ； Secondary data ； Threshold

參考文獻

吳家興, 林瑞雄, 謝貴雄, 邱文達, 陳麗美, 邱淑媞, . . . 蕭慧娟. (1998). 台灣北部國中學生氣喘盛行率調查. 中華公共衛生雜誌, 17(3), 214-225.

蘇有村, 陳國俊, 吳展耀, 翁慧卿, 陳施如, 劉惠君, . . . 牛震廣. (2008). 全民健康保險氣喘醫療給付改善方案改善多數兒童氣喘疾病嚴重度. [Government-sponsored Disease Management Program Improves Severity in the Majority of Pediatric Asthma in Taiwan]. 中華民國兒童胸腔醫學會雜誌, 5(4), 1-10.

Firoozi, F., Lemiere, C., Beauchesne, M. F., Forget, A., & Blais, L. (2007). Development and validation of database indexes of asthma severity and control. Thorax, 62(7), 581-587. doi:10.1136/thx.2006.061572

Jacob, C., Haas, J. S., Bechtel, B., Kardos, P., & Braun, S. (2016). Assessing asthma severity based on claims data: a systematic review. Eur J Health Econ. doi:10.1007/s10198-016-0769-2

Leidy, N. K., Paramore, L. C., Watrous, M., Doyle, J., & Zeiger, R. S. PEN3: DEVELOPMENT OF AN ALGORITHM FOR ESTIMATING ASTHMA SEVERITY FROM AN ADMINISTRATIVE COST DATABASE. Value in Health, 2(5), 394-395. doi:10.1016/S1098-3015(10)75867-8