Evaluation of a new ambulatory blood pressure device

We assessed the OSCILL-IT ambulatory blood pressure (BP) recorder (FIGI sri , Rome, Italy) according to the performance criteria set out by the British Hypertension Society (BHS) protocol. The OSCILL-IT is a portable, noninvas ive recorder that uses a process that correlates systolic, mean, and diasto lic areas, identified on the oscillations, to the cuff absolute pressure. A ccording to the recommendations of BHS, a large heterogeneous population (1 00 subjects: 52 men aged from 19 to 79-median 44 and 48 women from 19 to 74 -median 54) was recruited in order to assess accuracy and to analyze, in ad dition, the effects of observer agreement and BP level on the observer-devi ce differences. With reference to BP level, we suggest also a new graphic a pproach. Four sets of sequential, same arm, comparative BP measurements wer e obtained, performed by the OSCILL-IT recorder and two skilled clinicians using a mercury column, for each subject. We used a linear combination for the statistical evaluations. We confirmed the observer agreement through th e frequency distribution of BP as a function of the observer and through th e differences between observers. We compared OSCILL-IT with sphygmomanometr ic readings: the differences were not significant. A visual inspection, wit h the addition of regression lines, showed that there were no variations in differences at the changing of BP level. The difference between observers and OSCILL-IT was 0.2 +/- 5.3 mm Hg and 0.2 +/- 5.8 mm Hg both for systolic BP (SBP) and diastolic BP (DBP). The level of agreement, according to BHS criteria, showed that 64% of all systolic and 70% of all diastolic readings obtained by the OSCILL-IT were within 5 mm Hg of the sphygmomanometric det erminations. Therefore, the grade is C for SEP, even if 93% of SEP and 95% of DBP obtained by the OSCILL-IT were within 10 mm Hg of the sphygmomanomet ric determinations. These analyses demonstrate that the OSCILL-IT satisfies the accuracy parameters and the additional linear regression yields graphi cs more immediate. (C) 1998 American Journal of Hypertension, Ltd.