Development of equations for converting random-zero to automated oscillometric blood pressure values

Objective This study aimed to collect data to compare blood pressure values between random-zero sphygmomanometers and automated oscillometric devices and generate equations to convert blood pressure values from one device to the other. Methods Omron HEM-907, a widely used automated oscillometric device in many epidemiologic surveys and cohort studies, was compared here with random-zero sphygmomanometers. Two hundred and one participants aged 40-79 years (37% men) were enrolled and randomly assigned to one of two groups with blood pressure measurement first taken by automated oscillometric devices or by random-zero sphygmomanometers. The study design enabled comparisons of blood pressure values between random-zero sphygmomanometers and two modes of this automated oscillometric device – automated and manual, and assessment of effects of measurement order on blood pressure values. Results Among all participants, mean blood pressure levels were lowest when measured with random-zero sphygmomanometers compared with both modes of automated oscillometric devices. Several variables, including age and gender, were found to contribute to the blood pressure differences between random-zero sphygmomanometers and automated oscillometric devices. Equations were developed using multiple linear regression after taking those variables into account to convert blood pressure values by random-zero sphygmomanometers to automated oscillometric devices. Conclusion Equations developed in this study could be used to compare blood pressure values between epidemiologic and clinical studies or identify shift of blood pressure distribution over time using different devices for blood pressure measurements.

81 phase out the use of mercury instruments and use non-mercury devices instead [6].
82 Thus, the RZS (or HgS) cannot be used in longitudinal studies anymore, including 83 the INTERMAP China Prospective (ICP) Study, and an alternative non-mercury 84 device must be used to measure BP in these studies.
85 Most recent studies have used automated oscillometric devices (AODs) for BP 86 measurements; these detect pulse wave oscillations and estimate BP values via 87 algorithms. Protocols were developed to validate these devices before 88 recommendations for use [5,7].
89 At the individual level, long-term BP change has been shown to predict CVD risk [8- 90 10], which can also be used as a primary outcome in prospective studies. At the 91 population level, consistent information is needed to understand elevated BP and its 92 temporal trend and how the BP distribution has shifted over time.  156 Extensive efforts were adopted to reduce measurement errors between observers.
157 The observers attended training sessions regularly and passed examinations to be 158 certified for using both devices according to the standardized international protocols 159 (e.g., Manual of Operations of the INTERMAP Study for RZS BP measurement).
160 Each observer had made adequate practice of BP measurements with feedback 161 from senior investigators before the calibration study began. During BP 162 measurement, the two devices were positioned separately and the two observers 163 took the measurements independently without any communication, so that each 164 observer could not know the BP readings by the other observer. The two observers 165 took turns using the RZS or Omron device. In addition, during data collection, a 166 senior investigator monitored the procedures of the observers and audited them 167 regularly to ensure that observers measured BP according to the standardized 168 international protocols and to assure the quality of the data. 175 Descriptive statistics for continuous (means and standard deviations, SDs) and 176 categorical variables (frequencies and percentages) were calculated and are 177 presented separately for participants by procedure (A and B) and t-tests for two 178 independent samples were used to evaluate the differences between the two groups.
179 Mean SBP and DBP by the OSM, the OMM, the RZS and the mean difference of 180 each device/mode are presented by procedure (A and B). Paired t-tests were used 181 to evaluate the differences between OSM (or OMM) and RZS BP values.
182 Potential factors accounting for of BP differences between the OSM and the RZS, 183 including gender, age, device order, and their interaction terms were explored.
184 The difference between the two devices was plotted against the mean for the two   230 observed between OSM and RZS methods (Fig 2a), and OSM and OMM methods 258 361 The BP differences between the two devices of the present study were greater 362 among men than women, and BP values by AOD were slightly lower than RZS in 363 women whereas BP values by AOD were higher in men. These results are 364 consistent with another BP calibration study which enrolled 1,729 participants in the