子宮頸癌為婦女常見的癌症之一，是目前台灣婦女排名第一的惡性腫瘤。而一般無論是醫生或病人都需要瞭解，在接受手術及治療後，癌症復發或死亡的風險為何。因此針對病人的個別狀況，做預後的風險評估有其重要性。本文可以分成三個部份。第一部份主要是利用雙樣本檢定討論鱗狀上皮癌與腺癌的病人，在不同風險因子之下是否有顯著地差異，並且分別計算復發或死亡的相對風險及勝率比，進行預後的風險估計。第二部份則是藉由巢狀邏輯斯迴歸模型，比較在不同腫瘤最長徑之下，基質侵犯與血管淋巴侵犯的機率，探討因子間的相關性，接著將上述三個因子當成解釋變數，利用邏輯斯迴歸模型，討論淋巴結轉移的發生機率。最後主要是討論有基質侵犯的患者，手術後復發或死亡的預後風險評估。

The cervical cancer occurs frequently for women. It is also the number one malignancy for women in Taiwan. In general, both doctors and cancer patients need to pay attention to the risk of recurrence or death after the surgery or adjuvant therapy. Therefore, the risk assessment of prognosis is very important for the cancer patients.
In this study, we discuss the problems from three aspects. Firstly, we use the two sample test to examine the differences on patients with squamous cell carcinoma or adenocarcinoma under different risk factors. In order to compare the risk of recurrence or death for the patients with squamous cell carcinoma or adenocarcinoma, we compute the relative risk and odds ratio of recurrence or death between them. Secondly, we employ a nested logistic regression model to explain the relationship of depth of cervical invasion as well as lymphovascular space invasion with respect to tumor size. Moreover, we use the logistic regression model to estimate the probabilities of lymph nodal metastasis with respect to tumor size, depth of cervical invasion and lymphovascular space invasion. In the end, we analyze to the risk assessment of prognosis for those patients with depth of cervical invasion.

論文審定書 i
誌謝 ii
摘要 iii
Abstract iv
1 前言 1
2 資料描述 2
2.1 資料來源 2
2.2 資料處理 2
2.3 變數介紹 2
2.3.1 術前資料 3
2.3.2 病理資料 4
2.3.3 術後資料 5
2.3.4 變數基本統計量 5
3 研究方法 7
3.1 獨立雙樣本檢定 7
3.1.1 獨立雙樣本 t 檢定 7
3.1.2 皮爾森卡方檢定 8
3.2 風險評估 8
3.2.1 相對風險 8
3.2.2 勝率比 9
3.3 邏輯斯迴歸模型 9
3.4 巢狀邏輯斯迴歸模型 10
3.5 存活分析 11
3.5.1 存活曲線 11
3.5.2 Cox 比例風險模型 11
4 研究結果 12
4.1 鱗狀上皮癌與腺癌患者的區別 12
4.1.1 雙樣本比較 12
4.1.2 預後風險分析 13
4.2 腫瘤最長徑與風險因子間的關係 16
4.2.1 巢狀邏輯斯迴歸模型配適結果 16
4.2.2 邏輯斯迴歸模型配適結果 19
4.3 子宮頸癌病人預後風險評估 23
5 結語 26
參考文獻 27
A 附錄 28

1. Agresti, A. (2002). Categorical data analysis, 2nd Edition, Wiley, New Jersey.
2. Feder, M., Cowell, A. J., Duffy, S. Q. and Shi, Weihua (2003). Fitting nested logit models to complex survey data. 2003 Joint Statistical Meetings-Section on Survey Research Methods, 1399-1402.
3. Klein, J. P. and Moeschberger, M. L. (2003). Survival analysis: techniques for censored and truncated data, 2nd Edition, Springer, New York.
4. Lau, Hei-Yu, Juang, Chi-Mou, Chen, Yi-Jen, Twu, Nae-Fang, Yen, Ming-Shyen and Chao, Kuan-Chong (2009). Aggressive characteristics of cervical cancer in young women in Taiwan. International Journal of Gynecology and Obstetrics, 107, 220-223.
5. Rutledgea, T. L., Kamellea, S. A., Tillmannsa, T. D., Goulda, N. S., Wrightb, J. D., Cohnb, D. E., Herzogb, T. J., Raderb, J. S., Golda, M. A., Johnsona, G. A., Walkera, J. L., Mannela, R. S. and McMeekin, D. S. (2004). A comparison of stages IB1 and IB2 cervical cancers treated with radical hysterectomy. Is size the real difference? Gynecologic Oncology, 95, 70-76.
6. Therneau, T. M. and Atkinson, E. J. (1997). A introduction to recursive partitioning using the rpart routine. Technical Report, Section of Biostatistics, Mayo Clinic, Rochester, 61, 1-52.
7. Ward, K. K., Shah, N. R., Saenz, C. C., McHale, M. T., Alvarez, E. A. and Plaxe, S. C. (2012). Changing demographics of cervical cancer in the United States (1973-2008). Gynecologic Oncology, 126, 330-333.
8. 子宮頸癌的簡介與分期 (曾志仁醫師網頁)
http://www.csh.org.tw/dr.tcj/educartion/gyn-cancer-a13.htm