Analysis of an Observational Study
MetadataShow full item record
This item's downloads: 340 (view details)
Cara Dooley, John Hinde, Harry Comber and John Newell (2011) Analysis of an Observational Study . In: Conesa, D, Forte, A, Lopez-Quilez, A and Munoz, F eds. 26th International Workshop on Statistical Modelling, Valencia
The study presented below aimed to compare survival of colorectal cancer patients against survival of a sub-population with a secondary disease, in ammatory bowel disease (IBD). The data were taken from a observational study, that is there was no explicit design. The study had many complications, but the most significant aspect was that the number of controls was much greater than the number of cases of interest. Some techniques are used to overcome these obstacles, including: matching of the dataset, to make the controls and cases as similar as possible at time of diagnosis, effectively retrospectively fi tting a design; weighting of the data, using both the propensity score and the number of similar patients found in matching.