Applied Statistical Methods
This sub-group of the Cancer Epidemiology Research Group explores novel methodologies to enhance our understanding of cancer aetiology, epidemiology and treatment.
Some of our work includes a focus on:
Pharmacoepidemiology
Pharmacoepidemiology is the study of the safety, utilization and effectiveness of medications in large populations in the real-world setting.
Within the Cancer Epidemiology Research Group, our work in the field of cancer pharmacoepidemiology aims to determine the effects of commonly used medications on the incidence of cancer, as well as survival outcomes in cancer patients. Additional research evaluates the utilization and safety of treatments, with the overall aim of generating robust evidence on the uses, benefits and harms of medications in cancer patients that will improve decisions made by policymakers, care providers and patients.
The team have extensive experience working with large electronic healthcare datasets including the Clinical Practice Research Datalink, Scottish Registries and SAIL Databank. We also have a large network of international collaborators including researchers from the Belgian Cancer Registry, McGill University and the University of Southern Denmark.
Current academic staff working in this field include:
Risk prediction modelling
Risk prediction modelling aims to use information about individuals to make useful predictions about how likely they are to develop a health outcome, in order to inform screening, lifestyle or treatment decisions.
Within the Cancer Epidemiology Research Group, our work in the field of cancer risk prediction aims to improve the early detection of cancer by using clinical risk factors, biomarkers or genes to identify individuals at a high risk of cancer, who may benefit most from cancer screening. Additional research evaluates whether the information in individuals diagnosed with cancer can be used to estimate prognostic outcomes in order to inform shared treatment decisions.
Survival analysis and causal inference
Causal inference methodology is a rapidly developing area of applied statistics for use in observational datasets. Many scientific questions about cancer patient survival can only be investigated using observational datasets, in particular population-based cancer registry data that minimises the selection bias that otherwise would arise due to the inclusion criteria of clinical-trials methodology. In addition, certain real-world questions, such as the impact of delays in the patient-pathway on the outcome, or under-treatment, cannot be studied in the artificial environment of a clinical research centre.
The more comprehensive capture of individual patient clinical information, through better linkage of diverse information sources, including cancer registry and multi-disciplinary team meetings, is generating the conditions whereby causal inference methodology can be applied credibly. The effective use of observational data in this way holds enormous promise for studying the under-investigated research questions outlined above, but also for providing a treatment evidence-based for patient subpopulations for whom a randomised trial would not be economically attractive to carry out.
Within the Cancer Epidemiology Research Group, research using causal inference methodology is being used to:
1) define and estimate under-treatment in cancer patients,
2) estimate how much of poorer survival in young colorectal cancer patients is mediated through being diagnosed at a more advanced stage, and
3), assessing the potential for estimating treatment effects on survival using comprehensive clinical-audit datasets of cancer patients.
Current staff working in this field:
Systematic reviews and meta-analyses
Systematic reviews attempt to identify, appraise and synthesize all the available evidence that meets pre-specified eligibility criteria to answer a specific research question. A meta-analysis is a statistical analysis method that combines the results of multiple research studies and is often conducted as part of a systematic review.
Within the Cancer Epidemiology Research Group, there is significant expertise in the conduct of systematic reviews and meta-analyses investigating risk factors for cancer development and progression. To date, exposures of interest have included lifestyle factors, common medication use, as well as molecular and genetic biomarkers. We also have experience in conducting systematic reviews and meta-analyses evaluating the incidence of premalignant conditions and risk of progression to cancer.
Our systematic reviews and meta-analyses utilise explicit, systematic methods to minimise bias and produce more reliable findings with the ultimate aim of better informing clinical and public health decision making around cancer prevention and control.
Current academic staff working in this field include: