This protocol is part of the REPEAT Initiative, a project which samples published research studies conducted using large healthcare data (such as electronic health record or administrative claims) and replicates them by applying the publically reported methods to the same data source as the original authors. The goal is to better understand what information is missing from public reporting that prevents replication of the published results. This project will evaluate how commonly a set of specific design and analysis decisions are or are not reported in publications as well as how lack of clarity in one or more decisions impacts ability to replicate study findings. Our results will inform future policies and guidelines for reporting on healthcare database research.
This protocol focuses on one sampled study: "Glucagon-like peptide-1 analogues and risk of breast cancer in women with type 2 diabetes: population based cohort study using the UK Clinical Practice Research Datalink" by Hicks and colleagues. The Hicks paper compares the risk of breast cancer in users of two types of drugs used to treat diabetes. We will replicate this study based on methods reported in the publication.
This objective of this protocol is to replicate the study: "Glucagon-like peptide-1 analogues and risk of breast cancer in women with type 2 diabetes: population based cohort study using the UK Clinical Practice Research Datalink" by Hicks et al based on methods reported in the publication and appendices. We have created a checklist of specific study implementation parameters based on a comprehensive catalogue outlined in a consensus paper endorsed by the International Society of Pharmacoepidemiology and the International Society of Pharmacoeconomics and Outcomes research. We will start by reviewing the paper to identify which parameters from the catalogue are reported. We will then replicate the study population and analyses based on the study design and implementation parameters extracted during review.
The Hicks paper compares increased risk of incident breast cancer between female type-2 diabetes users of glucagon-like-peptide (GLP-1) and dipeptidylpeptidase-4 (DPP-4) from 2007 to 2015. We will focus on replicating the hazard ratio comparing the increased risk of incident breast cancer between GLP-1 and DPP-4 users over this period. The reference group was considered users of DPP-4 inhibitors. One year after entering the cohort, study participants were followed until their first breast cancer diagnosis. Descriptive statistics were calculated for both exposure groups, including incidence of breast cancer with 95% confidence intervals. Risk of incident breast cancer will be compared using a hazard ratio calculated from a time dependent Cox proportional hazards model. The model adjusted for a list of potential confounders, including year of cohort entry, age, body mass index, smoking status, alcohol-related disorders, haemoglobin A1c, duration of treated diabetes, and presence of microvascular complications of diabetes.
Health Outcomes to be Measured:
Use of GLP-1 analogues compared with use of DPP-4 inhibitors
Incident breast cancer