Readers should be aware that are labelling errors in Table 1, and in the text accompanying the description thereof. The table is based on the paper by Reich et al., which presented the results for a network meta-analysis (NMA). As the NMA used a Bayesian approach, the results in Table 1 represents the estimated probabilities for response (not mean % of patients who respond) and correspondening 95% Credible Intervals (not confidence intervals).
It would have been more accurate for Table 1 to be written with the labels as per the Reich et al. publication , and for the text to mention that it was a NMA that used a Baysian approach (as apposed to a frequentist approach).
Labelling errors in Table 1 and text
24 June 2014
Readers should be aware that are labelling errors in Table 1, and in the text accompanying the description thereof. The table is based on the paper by Reich et al., which presented the results for a network meta-analysis (NMA). As the NMA used a Bayesian approach, the results in Table 1 represents the estimated probabilities for response (not mean % of patients who respond) and correspondening 95% Credible Intervals (not confidence intervals).
It would have been more accurate for Table 1 to be written with the labels as per the Reich et al. publication , and for the text to mention that it was a NMA that used a Baysian approach (as apposed to a frequentist approach).
Competing interests
None declared