Increasing the validity and efficiency of blood pressure estimates using ambulatory and clinic measurements and modern missing data methods.

BACKGROUND Ambulatory blood pressure monitoring (ABPM) is considered the gold standard for BP measurement, compared to clinic BP measurements (CBP), which are a less valid predictor of target organ damage and cardiovascular events. However, ABPM is considerably more expensive than CBP, leaving BP researchers with a difficult dilemma: Use the less efficient CBP measure, or bear the cost of the more expensive ABPM. Recent developments in missing data methods, notably the two-method measurement (TMM) design, address this problem. With the TMM design, all research participants receive the less expensive CBP measure, but only a random subset receives the more expensive ABPM. The total number of participants must be increased, with additional participants receiving only CBP measurements. Even so, the TMM still reduces costs. METHODS We applied the TMM approach, which makes use of a "bias correction" structural equation model, to an empirical data set in which data were available for ABPM and CBP, as well as an echocardiographic measure of left ventricular mass (LVM). RESULTS Based on an estimated fivefold difference in cost for using ABPM compared to CBP, we found that statistical power can be considerably increased, or that BP measurement costs can be considerably reduced, when using this planned missing data design. CONCLUSIONS These benefits were observed with no loss of predictive validity (i.e., the observed association between BP and LVM). This suggests that the TMM design is a promising technique that in some studies may be able to decrease costs and/or increase one's power to detect effects.