Reproducibility or variability of casual and ambulatory blood pressure data: implications for clinical trials

Lack of reproducibility in a blood pressure measurement leads to a systematic underestimate of risk when only a few highly variable blood pressure estimates are taken. This has been called the 'regression dilution bias'. To determine the real risk produced by a lifelong blood pressure elevation requires many readings in order to estimate 'usual' blood pressure. Multiple readings improve the precision of the estimate of usual blood pressure. The 1-month variability in clinic and in 1, 2, 3, 4, 8 and greater than 20 ambulatory diastolic blood pressure values, measured in 100 untreated hypertensives, showed that whether the variability is measured as the standard deviation of the difference or as intermonth autocorrelations, the precision is improved by increasing numbers of blood pressure readings. The risks calculated from an elevation in blood pressure would have been up to 69% greater had ambulatory diastolic blood pressure been used to classify subjects in epidemiological studies. Repeat epidemiological investigations of the risks posed by certain levels of ambulatory blood pressure are needed.