根據行政院衛生署2009年統計，膀胱癌為台灣地區最常見的泌尿系統癌症且為國人主要癌症死亡排名第十三位。表淺型膀胱癌多為低惡性度癌症，大部分可以經尿道切除術切除，並輔以膀胱內灌注法預防腫瘤復發。我們感興趣的是膀胱癌輔以膀胱內灌注Thiotepa對於膀胱癌病患復發事件之療效，本篇使用退伍軍人管理局合作泌尿學研究團隊86個膀胱癌病患資料，為研究膀胱內灌注Thiotepa療程，對於膀胱癌病患復發的次數以及存活時間的影響，探討比較多維事件存活時間的三種邊際模型（marginal model）：AG模型、PWP模型、WLW模型與脆弱模型（frailty model）。

According to the Department of Health statistics in 2007, Bladder cancer is the most common genitourinary tumor ranked at the thirteenth important cancer in Taiwan. Most tumors of superficial bladder cancer are low grade cancers which can be removed by transurethral resection, supplemented by intravesical therapy in order to prevent the recurrence. We are interested in the treatment effects of thiotepa to the 86 bladder cancer patients’ recurrence from Vetrans Administration cooperative urological research group. To investigate this research problem, we focus on three marginal models (AG model, WLW model, and PWP model) and frailty models approaches of multivariate survival data analysis. In addition to studying the effect of thiotepa to bladder cancer patients’ recurrence and survival times under different models, we compare the performance of these approaches as well.

[1]民國96年癌症登記報告（2007）。中華民國行政院衛生署國民健康局。
[2] 林建甫（2008）存活分析。台北：雙葉書廊。
[3] Aalen, O. O. (1994). Effects of frailty in survival analysis. Stat. Methods Med. Res.,3, 227–2430.
[4] Abraham, T.K., James E. M., John A. F., Joseph, A. S., Richard, F. L, Randall, G. R. (1999). Report on the management of non-muscle-invasive bladder cancer (stages Ta, T1 and Tis), American Urological Association Inc.
[5] Aisbett, C. W. and McGilchrist, C. A. (1991). Regression with frailty in survival analysis. Biometrics, 47, 461–466.
[6] Andersen, P. K. and Gill, R. D. (1982). Cox’s regression model for counting processes: A large sample study. Annals of Statistics, 10, 1100–1120.
[7] Andersen, R. K., Gill, R. D., Nielsen, G. G. and S rensen, T. I. A. (1992). A counting process approach to maximum likelihood estimation in frailty models. Scandinavian Journal of Statistics, 19, 25–43.
[8] Badalament, R.A., Herr, H.W., Wong, G.Y., Gnecco, C., Pinsky, C.M., Whitmore Jr, W.F., Fair, W.R., Oettgen, H.F., (1987). A prospective randomized trial of maintenance versus nonmaintenance intravesical bacillus Calmette-Guerin therapy of superficial bladder cancer. Journal of Clinical Oncology, 5, 441-449.
[9] Boef, S. D. and Box-Steffensmeier, J. M. (2006). Repeated events survival models:The conditional frailty model. Statistics in Medicine, 25(20), 3518–3533.
[10] Cai, J., Cligg, L. X. and Sen, P. K. (1999). A marginal mixed baseline hazards model for multivariate failure time data. Biometrics, 55, 805-812.
[11] Chang, C.C., Cheng, H.L., Huang, K.H., Lin, S.H., Ling, Y.M., Tzai, T.S., Tong, Y.C., Yang, W.H. (1996). Postoperative Adjuvant Intravesical Instillation Therapy with BCG, Epirubicin and Thiotepa in Superficial Transitional Cell Carcinoma of Urinary Bladder. Journal of the Urological Association of R.O.C., 7, 187-192.
[12] Chen, J., Chiang, W.H., Chiu, T.Y., How, S.W., Hsieh, T.S., Hsu, T.C., Lin, F.S., Tsai, T.C. (1993). Clinico-patholigical study of bladder tumor. Journal of the Urological Association of R.O.C., 4, 1064-1070.
[13] Chiang, H.S., Guo, H.R. (1993). Geographical distribution of high-risk areas for bladder cancer in blackfoot disease endemic areas of Taiwan. Journal of the Urological Association of R.O.C.4, 1079-1085.
[14] Clayton, D. G. (1978). A model for association in bivariate life tables and its application in epidemiological studies of chronic disease incidence. Biometrika, 65,141–151.
[15] Cook, R. J. and Lawless, J. F. (2002). Analysis of repeated events. Statistical Methods in Medical Research, 11, 141–166.
[16] Cook, R. J. and Jerald, F. L. (2006). The statistical analysis of recurrent events.Springer.
[17] Cox, D. R. (1972). Regression models and life-tables (with discussion). Journal of the Royal Statistical Society, B–34, 187–200.
[18] Fleming, T. R. and Harrington, D. P. (1991). Counting processes and survival analysis.Wiley, New York.
[19] Grambsch, P. M. and Therneau, T. M. (2000). Modeling survival data:extending the Cox Model . Springer.
[20] Guo, G. and Rodriguez, G. (1992). Estimating a multivariate proportional hazards model for clustered data using the EM algorithm with an application to child survival in guatemala. Journal of American Statistical Association, 87, 969–976.
[21] Haskell (2001). Cancer Treatment. Saunders.
[22] Hougarrd, P.（1986a）. Survival models for heterogeneous populations derived from stable distributions. Biometrics, 73, 671-678.
[23] Hougaard, P. (1986b). A class of multivariate failure time distributions. Biometrics, 73, 387-396.
[24] Huber, P. J. (1967). The behaviour of maximum likelihood estimates under nonstandard conditions. Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability, 1, 221–233.
[25] Kaplan, E. L. and Meier, P. (1958). Non-parameteric estimation from incomplete observtion. Journal of American Statistical Association, 53, 457–481.
[26] Klein, J.P. (1992). Semiparametirc estimation of random effects using the cox model based on the em algorithm. Biometrics, 48, 798–806.
[27] Lawless, J. F. and Nadeau, C. (1995). Some simple robust methods for the analysis of recurrent events. Technometrics, 37, 158–168.
[28] Lin, D. Y. (1994). Cox regression analysis of multivariate failure time data: the marginal approach. Statistics in Medicine, 13, 2233–2247.
[29] Lin, D. Y., Wei, L. J. and Weissfeld, L. (1989). Regression analysis of multivariate incomplete failure time data by modeling marginal distributions. Journal of American Statistical ssociation, 84, 1065–1073.
[30] Lin, D. Y. and Wei, L. J. (1989). The robust inference for the cox proportional hazard model. Journal of the American Statistical Association, 84, 1074–1078.
[31] Lin, D. Y., Wei, L. J., Yang, I., and Ying, Z. (2000). Semiparametric regression for the mean and rate functions of recurrent events. Journal of the Royal Statistical Society, B–62, 711–730.
[32] Manton, K. G., Stallard, E. and Vaupel, J. W. (1979). The impact of heterogeneity in individual frailty on the dynamics of mortality. Demography, 16, 439–454.
[33] Netto, R. R. Jr., Lemos, G.C. (1983). A comparison of treatment methods for the prophylaxis of recurrent superficial bladder tumors. Journal of the Urology, 129, 33-34.
[34] Oakes, D. (1992). Frailty models for multiple event times. Survival analysis: state of the art, pp. 371–379.
[35] Peterson, A. V., Prentice, P. L. and Williams, B. J. (1981). On the regression analysis of multivariate failure time data. Biometrika, 68, 373–379.
[36] Philip Hougaard. (2000). Analysis of multivariate survival data.Springer.
[37] Von der Maase H, Hansen, S. W., Dogliotti, L., Oliver, T., Moore, M. J. (2000). Gemcitabine and cisplatin versus methotrexate, vinblastine, doxorubicin, and cisplatin in advanced or metastatic bladder cancer: results of a large, randomized, multinational, multicenter, phase Ⅲ study. Journal of clinical oncology, 18, 3068-3077.