Comparative effectiveness studies assess the benefits and risks of newer and older treatment using observational studies. There are rising concerns of combining new (those who have never used the older and newer therapies) and prior treatment users (those with a prior history of using older and newer therapies) in comparative effectiveness studies. Existing methods propose to only include new treatment users but this would exclude a large patient population in the real world. A more inclusive method is to match patients of newer to older therapies on prior history of treatment but this method is infeasible when there is increasing trend of using the newer compared with older treatment. Therefore, the objective of the study is to explore two alternative methods to include all patients in the real world. Specifically, in the first approach we will randomly select a patient group to make the older treatment group more comparable to the newer treatment group and in the second approach we will consider a scenario where patients change their treatment over time. The results produced from this research will provide invaluable evidence to researchers about the different methods that can suitably account for patients with a prior history of treatment in future studies.
Currently, there are concerns of prevalent user bias in comparative effectiveness studies but current methods to mitigate this bias have limitations. Therefore, the purpose of the study is to explore alternative study design and analytical techniques to account for patients with a prior history of treatment in a comparative effectiveness study of insulin glargine and breast cancer incidence. A cohort of women 40 years or older with at least one prescription for any insulin from 1988-2012 will be created, with follow-up until February 28, 2015. Two approaches, which will account for women with a prior history of insulin use, will be compared. First, a pseudo-matching approach will be conducted by randomly selecting prescription dates among prior Neutral Protamine Hagedorn (NPH) insulin users (comparator). Second, a time-dependent approach will be used where women are considered unexposed until the first insulin glargine prescription and exposed until end of follow-up. We will compare insulin glargine with NPH because these two insulins are recommended by NICE guidelines as last line treatment options for patients with type 2 diabetes. The Cox proportional hazards model will be used to estimate the adjusted hazard ratios and 95% confidence intervals for each of the approaches.
For both the pseudo-matching and time-dependent approach, we will assess the effect of insulin glargine on breast cancer. First, primary diagnosis of breast cancer will be defined by Read codes. Breast neoplasms will be classified by Read codes starting with 'B'.
Samy Suissa - Chief Investigator - Sir Mortimer B Davis Jewish General Hospital
Samy Suissa - Corresponding Applicant - Sir Mortimer B Davis Jewish General Hospital
Jennifer Wu - Collaborator - McGill University
Laurent Azoulay - Collaborator - McGill University