Grant: Discovery Grants Program - Individual
University of Ottawa
$16,000.00
Analysis of biased data arising from cross-sectional sampling
EF 2010-11
Résumé
In the analysis of time-to-event data, when interest lies in measuring the distribution of the time from an initiating event, such as the onset of a disease, until an event of interest, e.g. death, it is often impractical or impossible to sample individuals from the time of the initiation of the condition of interest. Thus researchers may recruit subjects through cross-sectional sampling, which is a data collection scheme where the subjects have already experienced onset of the condition of interest prior to recruitment. The time-to-event of these individuals, known as prevalent cases in epidemiology, tends to be longer than that in the general population of individuals with the condition of interest (the incident cases), thus the data on durations obtained through cross-sectional sampling are biased compared to the population of interest, as individuals with longer event-times are more likely to be recruited into the sample.
Data from prevalent cases have particular features that prevent direct extension of the standard methodologies in survival analysis which rely on incident cases. One such feature is informative censoring, as independence of censoring and failure times is one of the necessary assumptions for the martingale theory behind the tools used for incident cases. Another feature is covariate bias. In standard regression settings, the marginal distribution of the covariates holds no information on the parameters of interest, thus analyses are carried out by conditioning on the observed covariate values. When event-times are sampled with a bias, the marginal distribution of the associated covariates also differ from that of the incident population and depends on the regression parameters, and methodologies that take a joint approach offer more efficient estimators for these parameters.
The main theme of this project is to train graduate students to study time-to-event data arising from cross-sectional sampling and propose appropriate methodologies for the analysis of such data. Each of these methodologies will be tested using simulation studies and applied to real data, such as data on survival with dementia collected as a part of a nation-wide study called Canadian Study of Health and Aging (CSHA).
Programme
Discovery Grants Program - Individual
Bénéficiaire
Emplacement
Ontario, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data
Tel que publié
Application id
371516-2009
Name nom
Bergeron, PierreJerome
Department département
Mathematics and Statistics
Organization id
28
Institution établissement
University of Ottawa
Province en
Ontario
Province fr
Ontario
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2010
Competition year année de concours
2009
Award amount
16000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
Discovery grants programs
Group fr
Programmes de subventions à la découverte
Committee code
14
Committee name en
Statistical Sciences
Committee name fr
Statistique
Area of application code
1211
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Medical and health sciences
Area of application fr
Sciences médicales et de la santé
Research subject code
3005
Research subject group en
Statistics and probability
Research subject group fr
Statistique et probabilité
Research subject en
Biostatistics
Research subject fr
Biostatistiques
Partie
2
Nb partie
5
Application title
Analysis of biased data arising from cross-sectional sampling
Keyword
NSERC
Application summary
In the analysis of time-to-event data, when interest lies in measuring the distribution of the time from an initiating event, such as the onset of a disease, until an event of interest, e.g. death, it is often impractical or impossible to sample individuals from the time of the initiation of the condition of interest. Thus researchers may recruit subjects through cross-sectional sampling, which is a data collection scheme where the subjects have already experienced onset of the condition of interest prior to recruitment. The time-to-event of these individuals, known as prevalent cases in epidemiology, tends to be longer than that in the general population of individuals with the condition of interest (the incident cases), thus the data on durations obtained through cross-sectional sampling are biased compared to the population of interest, as individuals with longer event-times are more likely to be recruited into the sample.
Data from prevalent cases have particular features that prevent direct extension of the standard methodologies in survival analysis which rely on incident cases. One such feature is informative censoring, as independence of censoring and failure times is one of the necessary assumptions for the martingale theory behind the tools used for incident cases. Another feature is covariate bias. In standard regression settings, the marginal distribution of the covariates holds no information on the parameters of interest, thus analyses are carried out by conditioning on the observed covariate values. When event-times are sampled with a bias, the marginal distribution of the associated covariates also differ from that of the incident population and depends on the regression parameters, and methodologies that take a joint approach offer more efficient estimators for these parameters.
The main theme of this project is to train graduate students to study time-to-event data arising from cross-sectional sampling and propose appropriate methodologies for the analysis of such data. Each of these methodologies will be tested using simulation studies and applied to real data, such as data on survival with dementia collected as a part of a nation-wide study called Canadian Study of Health and Aging (CSHA).