SRVL DTA ANLYS EPDM
SRVL DTA ANLYS EPDM EPI 537
Popular in Course
Popular in Epidemiology
This 26 page Class Notes was uploaded by Garry Marvin on Wednesday September 9, 2015. The Class Notes belongs to EPI 537 at University of Washington taught by Staff in Fall. Since its upload, it has received 31 views. For similar materials see /class/191974/epi-537-university-of-washington in Epidemiology at University of Washington.
Reviews for SRVL DTA ANLYS EPDM
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 09/09/15
BiostatistiosEpidemiology 537 Survival Data Analysis in Epidemiology Professor Norman Breslow Department of Biostatistios University of Washington Winter Quarter 2008 Lecture 1 Biost Epi 537 Censored Survival Data Year of Year of Reason Survival Subject Entry Exit for Exit Time 1 20000 20014 D 14 2 20002 20030 A 28 3 20004 20008 D 04 4 20006 20024 D 18 5 20008 20030 A 22 6 20010 20022 D 12 7 20012 20030 A 18 8 20014 20030 A 16 9 20016 20024 D 08 10 20018 20030 A 12 DDied AAive at study end censored Winter 2008 Lecture 1 Censored Survival Data Subject 2000 2001 2002 2003 Calendar Time BioslEpi 537 Winter 2008 Lecture 1 3 Workpage Biost Epi 537 Winter 2008 Lecture 1 Subject BioslEpi 537 New Origin for Time Scale Entry on Study O 1 2 3 Time on Study Winter 2008 Lecture 1 5 Choice of Time Scale Scale Origin Comment Calendar time Birth of JC Rarely meaningful Study time Diagnosis or treatment Clinical trials Study time First exposure Occupational epidemiology Age Birth subject Often best for epidemiology Biost Epi 537 Winter 2008 Lecture 1 Different Types of Censorship Subject 23456789 1 2000 2001 2002 2003 Calendar Time Same as before except two subjects now withdrawn W from observation lost to followup before closing date 2003 Biost Epi 537 Winter 2008 Lecture 1 TimetoEvent Data Y0 year of entry Y1 year of exit earliest of D death or event failure A administrative censoring end of study W withdrawal from observation before event 0 moved away LTFU T Y1 Y0 timetoevent for those who have event for those who don t Observe data censored timetoevent T minT C T time under observation 1 deadfailed 5 O censored so T T when 6 1 whereas T C timetocensorship othenlvise where T denotes underlying true timetoevent Biost Epi 537 Winter 2008 Lecture 1 8 TimeonStudy including Withd rawals Hi Subject i 4 i U 4 o o o 0 o in N o N 0 Time on Study BiostEpi 537 Winter 2008 Lecture 1 9 Critical Assumption of Independence Assume For subjects censored at fixed c probability of survival past some future time t gt c same as probability of survival past t for subjects who were merely known to have survived past c In symbols PrT gt tT c6 o PrT gt tT gt c Subjects censored at C are no different in terms of future survival When does independence fail 0 Administrative censoring o Losstofollowup dropouts Biost Epi 537 Winter 2008 Lecture 1 10 TimetoEvent Data in Epidemiology Unlike death the event of interest in epidemiologic studies is not always certain may not occur 0 death from lung cancer 0 incidence of Type II diabetes Implications o T not defined for subjects that do not experience the event 0 St PrT gt t is not always meaningful But timespecific event rate instantaneous hazard 1 At Prt g T lt t AtT Z t still at risk at t is meaningful and we will learn to make inferences about it o Often t is age and Mt is disease incidence rate at age t Biost Epi 537 Winter 2008 Lecture 1 11 Example Chemotherapy for Rhabdomyosarcoma Children s Cancer Group CCG Study 603 a Randomized trial for children with localized completely resected disease 0 15 randomized to surgery only control Part IA 0 17 randomized to surgery plus chemo treated Part IB o Observational study for children with microscropic residual disease 0 11 treated with surgery plus chemo Part A o Is time to first recurrence affected by chemotherapy Heyn RM Holland R Newton WA Jr Tefft M Breslow N Hartman JR The role of combined chemotherapy in the treatment of rhabdo myosarcoma in children Cancer 342128 41 1974 Biost Epi 537 Winter 2008 Lecture 1 CCG603 Clinical Course of Patients in Control Group IA xrRAv ACT 0 vcn o nzcunmvcz R w Mm DEATH A lewe and WELL cw CYELDPHDSPHAMDE TONGUE THVGH AND GROIN cuzsr WALL BLADDER EXP wimouY TUMOR msumm AREA DIAGNOSIS BiostEpi 537 Winter 2008 Lecture 1 13 CCG603 Clinical Course of Patients in Treated Group IB x RAV c A r u m NECK i RE RRENCEiRorMETASTASISWJ g DEAYH CHEST WALL A UVWE and WELL cm tympanime Pzalrzsmumn cussr wan uzcx METAYARSAL BiostEpi 537 Winter 2008 Lecture 1 BiostEpi CCG603 Clinical Course of Patients in Group IIB microscopic residual XRAY I Mr D o VCR o METASTASS 1M ORBIT 5 EA u LIVING and WELL BUTTOCK PERIANAL AED WALL CALF LARYNX FROM 537 Winter 2008 Lecture 1 Example Atomic Bombing Survivors RERF Breslow amp Day II Appendix IB 40 50 Subject 30 1950 1955 1960 1965 Calendar Year Left truncation in 1950 0 Subjects who die earlier never enter cohort Biost Epi 537 Winter 2008 Lecture 1 16 Atomic Bombing Survivors RERF o Is estimated dose of radiation related to the risk of death from o leukemia o stomach cancer 0 any cause 0 Cohort 163720 survivors of Hiroshima and Nagasaki bombings listed in 1950 censuses o T months since exposure bomb 0 Event death from specific cause 0 Covariates O 33D 0 33A age in years ATB 0 Q Adjusted for 33A is the risk of leukemia death related to xD o AtxD 33A risk of leukemia death at month t since exposure given 33D13A Biost Epi 537 Winter 2008 Lecture 1 Example Welsh Nickel Refiners Cohort Breslow amp Day II Appendix ID 60 Subject 20 1920 1940 1960 1980 Calendar Year Cohort all male workers in South Wales nickel refinery on at least 2 paysheets 1st week oprril 1929 1934 1939 1944 1949 Biost Epi 537 Winter 2008 Lecture 1 18 Welsh Nickel Refiners Cohort Q How is the risk of lung and nasal sinus cancer associated with nickel refining o How does age at exposure relate to risk 0 How does cessation of exposure relate to risk Cohort all male workers on at least 2 paysheets 1st week of April 1929 1934 1939 1944 1949 in a South Wales nickel refinery o n 968 18 lost to followup LTFU o Followed through 1981 0 High exposure priorto 1925 T years since first employed Event death from nasal sinus cancer may not occur Biost Epi 537 Winter 2008 Lecture 1 Example Recurrence in Patients with Herpes V Subject O 2 4 6 8 10 12 Time since end of initial episode months Cohort attendees at a herpes clinic most with primary lesions followed 1 year Biost Epi 537 Winter 2008 Lecture 1 20 Recurrence in Patients with Herpes Q1 Is the risk of recurrence rate related to 0 treatment with Acyclovir 0 characteristics of the primary episode x5 severity score for initial episode 33D duration of initial episode 02 Is x5 related to time to first recurrence Q3 ls time to first recurrence related to time from first recurrence to second recur rence o A1txD x5 risk of first recurrence at day t given 33D x5 0 A20 T1xD x5 T1 risk of second recurrence t T1 days after first recurrence given 33D x5 T1 Biost Epi 537 Winter 2008 Lecture 1 21 Example Breast Cancer in Young Women 0 mquot C i C 6 C 2 g O A U 0 BC O arc ABC Oio C i i i i i O 10 20 30 40 Age years Cohort conceptual All women born after 1965 in King Pierce or Snohomish Counties BiostEpi 537 Winter 2008 Lecture 1 22 Example Breast Cancer in Young Women Followed through 2005 for reproductive history benign breast disease BBD and occurrence of breast cancer Q How is reproductive history and benign breast disease related to breast cancer risk T age in years Event occurrence of primary BC may not happen Timedependent explanatory variables 0 X1t parity at age t o X2t 1 if diagnosis of BBD by age t 0 othenvise Q Are X1t X2t related to the risk of BD at age t AtX1t X2t rate of BC at age t for a woman with parity X1t and prior BBD status X2 t Biost Epi 537 Winter 2008 Lecture 1 23 The Lexis Diagram 07 IA gt A 1 gt N 3 IA 393 U IA a C I 0 IA 0 o 1 E r7 C uJ 39 I l l D l I 20 00 20 01 20 02 2003 Calendar Time 0 Two time scales 0 Calendar time 0 Time on study BiostEpi 537 Winter 2008 Lecture 1 24 The Lexis Diagram ll Age years 50 55 45 40 1950 1955 1960 1965 1970 Calendar Year 0 Disease incidence by age and calendar year BiostEpi 537 Winter 2008 Lecture 1 25 Critical Issues in Survival Data Analysis 0 Choice of time variable may be multiple 0 Time on study 0 Age 0 Right censorship is it independent of timetoevent 0 End of study administrative censorship 0 Loss to followup moved dropouts o Competing risks other causes of death coming 0 Left truncation staggered entry 0 Don t observe entire history from time 0 o Cohort definition must never depend on events that occur after start of FU 0 Severe selection bias othenvise Biost Epi 537 Winter 2008
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'