Reproducibility of ambulatory blood pressure monitoring

OBJECTIVE: To investigate whether blood pressure monitoring is reproducible. DESIGN: Reproducibility of ambulatory blood pressure monitoring data was assessed by means of traditional and relatively new statistical methods, namely correlation coefficients, regression analysis and agreement analysis. METHODS: Ninety-one normotensive and hypertensive, uncomplicated outpatients underwent monitoring twice (mean interval 241 days). Data were analysed for reproducibility, correlation, difference and adaptation. Analyses were performed to verify the reproducibilities of the diagnosis of hypertension and of the treatment assessment. RESULTS: Ambulatory blood pressure monitoring is highly reproducible in terms of traditional statistics, but not in terms of the agreement analysis (error as high as 18 mmHg), although it performs better than does office sphygmomanometry (error as high as 38 mmHg). Reproducibility is acceptable in normotensive subjects and in patients who respond to treatment, but untreated hypertensives and those who do not respond to treatment show a worse ratio. CONCLUSION: The reproducibility of ambulatory blood pressure monitoring requires improvement. We do not know how many repeated measurements we need before diagnosing hypertension. Spontaneous variability of blood pressure interferes with blood pressure reproducibility. Diagnosis and treatment assessment in hypertension must take into account poor reproducibility.