Is there an observer bias with random-zero blood pressure machines?

BACKGROUND: Random Zero (R-Z) sphygmomanometers are used in many epidemiological studies and clinical trials. A recent study reported that blood pressure measured with an R-Z sphygmomanometer was significantly correlated with R-Z values. In addition, the investigators showed that such blood pressure measurements were on average a few mmHg lower than those obtained with standard mercury column devices. It was suggested that a major part of this discrepancy could be explained by incorrect operation of the R-Z machine. DESIGN: This was an observational study based on 5021 blood pressure measurements that had been collected with an electronic machine that permits positive and negative random-zero values. METHOD: The association between corrected blood pressure and random-zero values was calculated separately for positive and negative offset values. RESULTS: The relationship between corrected blood pressure and random-zero values measured with the electronic device was similar to that observed when the R-Z machine was used correctly. However, we found that this correlation occurred primarily for high random-zero values. CONCLUSION: We attribute our findings to observer reluctance to accept high blood pressure values at the first Korotkoff sound and suggest that indirect blood pressure measurements with high values are on average biased towards lower values, regardless of the instrument used. The effect may, however, be more pronounced with R-Z machines because they display higher blood pressure values. The magnitude of the bias of 2-3 mmHg, which appears to be limited to high blood pressure values measured at high random-zero values, may not be large enough to outweigh the advantages of R-Z devices, especially in clinical trials and epidemiological studies.