Kappa statistic

As a teacher of basic skills in evidence-based medicine (EBM), I have appreciated the CMAJ articles that have been appearing in the EBM “tips” series. In particular, I was happy to see the discussion of the kappa statistic,1 specifically the calculation of chance agreement (e.g., Table 3 in the article).

Unfortunately, discussions of kappa tend to focus on dichotomous variables, such as positive or negative results on mammography or the presence or absence of Murphy's sign. In cases of continuous variables, such as jugular venous pressure, how is agreement compared and kappa calculated?