Wasn’t the same cohort used to validate both models? If so shouldn’t each decile of the same cohort have the same observed risk in both graphs? Wouldn’t only the predicted risk change (based on whether Qstroke or Framingham is used?) Since the observed risk changed I thought it was depicting different deciles of predicted risk (1 low and 10 high) so this clue confused me. Can you please clarify further.
Hi --
Because predicted risks differ, the people making up each decile of predicted risk will not be the same between the Qtroke and Framingham scores. For example, the 10% of the population in the bottom decile of the Q-stroke risk won't be the same people as those in the bottom decile of the Framingham risk score.
Tom
Sorry Dr. Newman, this hint also has me confused for different reasons. I see the scales of the 2 graphs are different ont the y axis (ten year risk%), but the x axis is deciles of predicted risk for both Qstroke and framinham risk right?
Yes, that is correct, but as I pointed out earlier, the people in each decile may differ for Qstroke and Framingham.