Objective: We describe the National Health and Nutrition Examination Survey (NHANES) blood pressure (BP) observer training and protocol standardization and evaluate the quality of BP measurement.
Methods: The participants were persons aged 8 years and older who had their BP measured (n=7467) during NHANES 1999-2000. Cuff width/arm circumference ratio (CW/AC), end digit preference, and observer agreement were examined.
Results: In stepwise principal components multiple regression analysis, CW/AC accounted for less than 2% of variability R(2) in all readings. The frequencies for all end digits were close to 20% ("0" end digit=21% systolic and 23% diastolic). No overall observer effect was present for mean systolic BP readings. A significant observer effect (P<.0001) was detected for mean diastolic BP readings of <90 mm Hg. For readings of > or =90 mm Hg, there was no significant observer effect (P=.157).
Conclusion: We conclude that NHANES BP measurements do not demonstrate the variability that is commonly caused by observer and technical error.