Journal Club 2

In the article, cases are defined as individuals with…

• 1. ... with an incident diagnosis of diabetes 

• 1. … with an incident diagnosis of renal cell carcinoma 

• 1. ... with a 2-year diagnosis of renal cell carcinoma

• 1.  … with a current treatment using metformin

Which of the following statements is true regarding the association between metformin use and RCC: 

• There is a positive association between metformin use and RCC when other variables are not considered

• There is a positive association between metformin use and RCC when adjusting for other variables

• Metformin is likely to cause RCC when the patients have >30 prescriptions

• Metformin is a protective factor for RCC development

Another feasible design for studying the relationship between medication and the incidence of RCC using already existing CPRD registry data would be:

• Randomised Clinical Trial

• Prospective cohort study

• Retrospective cohort study

• Cross-sectional study

Controls are selected by recruiting them from the same General Practice as cases. According to the authors, the reason for this decision is:

• Increasing recruitment feasibility

• Controlling for socioeconomic variables

• Controlling for opportunity bias

• Increasing statistical power to achieve 1:6 allocation ratio

The study considers that cases were positive two years before the diagnosis (index date) because of disease latency. Among other considerations, controls are matched with cases according to negative diagnosis at the same index date. Do you think this decision involves any threat to results? What is a possible bias introduced by this decision? Explain why.