1

I was trying to replicate an example from Knol et al (Knol 2012, Int J Epidemiol, Recommendations for presenting analyses of effect modification and interaction) but I dont get the same results as the authors, who give the following example data (part of table 2):

In the unexposed group (chronic disease score = 0) there are 243 cases (diabetes) and 24195 controls (no diabetes).

In the exposed group (chronic disease score > 0) there are 338 cases (diabetes) and 11260 controls (no diabetes).

So the risk in exposed = 338/(338+11260) = 0.029

Risk in unexposed = 243/(243+24195) = 0.0099

RR = 0.029 / 0.0099 = 2.9

Yet the authors provide an RR of 1.55 and I cannot grasp how they arrived at that figure.

Erik
  • 25
  • The key lies in the caption of Table 2 "RRs are adjusted for age, sex and benzodiazepine use." – mdewey Dec 11 '17 at 14:24
  • Thanks! That seems to be it. Too bad this makes the analysis impossible to replicate... – Erik Dec 11 '17 at 14:35

1 Answers1

1

Accoding to the comment by mdewey: The key lies in the caption of Table 2 "RRs are adjusted for age, sex and benzodiazepine use."

Erik
  • 25