A meta-analytical approach to the pooling of blood pressure values in children.

Meta-analysis of blood pressure (BP) studies in children was performed in order to evaluate whether BP readings obtained in various studies may be pooled. In total 30 original reports on BP in children were analyzed and the age and sex-specific BP values reported for 12 and 15-year-old subjects in 11 of these were tested for homogeneity. The differences in sample size-weighted means between the devices used (with 95% confidence intervals) were evaluated. The BP values reported in different studies for systolic Korotkoffs first phase BP (K1) and diastolic Korotkoffs fourth (K4) and fifth (K5) phase BP were shown to be too heterogeneous (p < 0.001) to be pooled. Heterogeneity and measurement biases in the BP results were caused by differences between the devices and the diastolic BP indicator (K4 vs K5) and by observer biases such as uncontrolled terminal digit preference. The device-specific weighted means for K1, K4 and K5 BP in 12 and 15-year-old subjects showed statistically significant differences between random zero (RZS) and ordinary mercury (OMS) sphygmomanometer (p < 0.001) the difference (OMS-RZS) being from 0.7 to 6.6 mmHg for K1 (p < 0.001, except in 12-year-old girls), from 1.1 to 7.1 mmHg for K4 (p < 0.001, except in 12-year-old girls) and from 4.5 to 7.9 mmHg for K5 (p < 0.001). Such a discrepancy was also apparent between ordinary and automated mercury sphygmomanometer (AMS) in the case of diastolic K4 and K5 BP (p < 0.001), but not in K1. BP results obtained in different investigations using different BP devices and Korotkoffs phases as indicator of diastolic BP should not be pooled. More widely accepted recommendations are needed to standardize the measurement of BP in children.