Download Recurrence Risk in PBC and SCC - Biostatistics & Epidemiology - Prof. Norman Breslow and more Assignments Epidemiology in PDF only on Docsity! Biostat/Epi 537 Winter Quarter, 2008 Professor Breslow HOMEWORK #4 (Due Thursday, February 7 in class) Reading: H&L: Chapter 4, §7.1,7.2 Articles: Prognosis in Primary Biliary Cirrhosis (Hepatalogy 10:1-7,1989) Local Recurrences and Distant Metastases ... (JNCI 87:19-27, 1995) available from the class web page. Homework: 1) Using the HERPES.DAT dataset considered in previous homeworks, analyze the association between the duration of the primary episode and the risk of recurrence as a function of time since the end of the primary episode. Using the relative (instantaneous) risk as the measure of association, perform both an unadjusted analysis and an adjusted analysis where the adjustment variables include age, gender, and factored versions of HSV type (3 levels) and acyclovir treatment group (4 levels). Look for evidence of interaction effects between duration of primary lesion and each of the adjustment variables. What do you conclude from these data? (Hint: Use STATA’s “xi” command in lieu of constructing dummy variables yourself.) 2) (Continued) Now fit the model with terms for duration, age, gender and all pairwise interactions involving these three variables. For interpretability, it would be a good idea to “center” duration and age at, say, 20 weeks and 25 years, respectively. a) Using the estimated coefficients, what is the rate ratio (RR) of recurrence for a 25 year old woman whose primary lesions took 60 weeks to heal compared with another 25 year old woman whose initial lesions took only 10 weeks to heal? b) What is the RR comparing 60 to 10 weeks duration for a 20 year old man? 3) (Continued) One reason longer durations of primary episode may be associated with shorter times to first recurrence (after the end of the primary episode) is that what really matters biologically is the time since the beginning of the primary lesion. With this in mind, refit the basic model described in problem #1 with a new time variable: “timefromdx” = time from diagnosis to first recurrence. a) Explain the change in the duration coefficient from that found in #1 above. b) What criticism do you have of this model? 4) On the class web page are data (ECOGdata.txt) and corresponding data description (ECOGdesc.txt) from a study of 194 patients with squamous cell carcinoma. The study was from a clinical trial conducted by the Eastern Cooperative Oncology Group. Patients were followed until failure defined as: local spread of disease (cause 1); or metastatic spread of disease (cause 2) { there were 67 subjects whose follow-up was censored. Three covariates are of interest: performance status measured as 0=ambulatory, and 1=non-ambulatory; treatment 0=treatment A, 1=treatment B; and age in years at start of the study.