Comparison of Exercise Blood Pressure Measured by Technician and an Automated System

Transcription

1 Clin. Cardiol. 7, (1984) Clinical Cardiology Publishing Co., nc. Practitioner's Corner Comparison of Exercise Blood Pressure Measured by Technician and an Automated System J. A. GARCA-GREGORY, M.D., A. S. JACKSON, P.E.D., J. SrUDEVLLE, M.D., W. G. SQURES, Ph.D., C. A. OWEN, M.D. Section of Cardiology, Kelsey-Seybold Clinic; Department of Health, Physical Education and Recreation, University of Houston; The Cardiopulmonary Laboratory, NASA/Johnson Space Center, Kelsey-Seybold Clinic, Houston, Texas, USA Summary: We evaluated the automated system Blood Pressure Measuring System (BPMS) developed by NASA on 277 adult males who elected to have a treadmill test as part of their annual physical. The BPMS uses acoustic transduction with a computer-assisted ECG gating to detect nonsynchronous noise. The BPMS readings were compared to pressures simultaneously measured by trained technicians. For all stages of work, BPMS readings were higher for systolic and lower for diastolic than technician readings. At peak stages of work, BPMS systolic pressures were about 20 mmhg higher than technician readings. Within each 3-min workstage, BPMS readings were found to be more inconsistent than technician readings. The standard errors of measurement for BPMS were from two to three times higher than technician values. These data showed automated blood pressure readings were significantly different than technician values and subject to more random fluctuations. These findings demonstrate the need to view exercise blood pressure measured by automated systems with caution. Address for reprints: Dr. J. A. Garcia-Gregory Section of Cardiology Kelsey-Seybold Clinic 6624 Fannin Houston, Texas 77030, USA Received: February, 1984 Accepted: February 10, 1984 Key words: exercise blood pressure, automated blood pressure measurement, exercise hypertension ntroduction Hemodynamic changes induced by exercise are of diagnostic value to practicing physicians. Walking and/or jogging on a treadmill introduces both arm movement and noise which makes it more difficult to measure blood pressure. For diagnostic and safety reasons, interest in measuring exercise blood pressure accurately is high and several electronic systems have been developed. Glasser and Ramsey (1981) have provided data to suggest that an automated system which used electrocardiographic gating and microcomputer processing of nonsynchronous noise to isolate acoustically the Korotkoff sounds measures blood pressure accurately. The correlations between exercise blood pressure measured by technicians and the automated system were high, but standard deviations for systolic pressure differences varied 7-9 mmhg, which showed that the readings varied substantially. Therefore, this study was initiated to evaluate a similar automated system developed by the National Aeronautics and Space Administration (NASA). The study was designed to determine: (1) if there were systematic differences in exercise blood pressure measured by a trained technician and the NASA automated system, and (2) if the automated readings were less subject to random fluctuations than readings by a trained technician.

2 316 Clio. Cardio!' Vo!' 7, May 1984 Methods The Blood Pressure Measuring System (BPMS) was designed to measure an astronaut's blood pressure during rest and at exercise on a bicycle ergometer. The BPMS measured blood pressure by the noninvasive Korotkoff sound technique (Nolte, 1977). To measure blood pressure, the occlusive cuff was automatically filled to one of three pressures: 160,200, or 250 mmhg. The occlusive cuff pressure was decreased at a systematic rate of 7 mmhg/s. A microphone with the necessary electrical modules was used to monitor the appearance and disappearance of the Korotkoff sounds and register systolic and diastolic pressures. The system was designed to coordinate Korotkoff sounds with the electrocardiogram. At each QRS complex of the heart, the electrocardiograph supplied a switch closure used to generate an "operating window," which was the time interval during which the system would accept the sounds from the microphone. A feature of BPMS was the capacity to record pressure electronically and simultaneously on a mercury manometer which was part of the BPMS. The mercury manometer was attached by a t connector to the gas line used to increase and decrease cuff pressure. The BPMS was used to measure exercise blood pressure during graded exercise stress tests administered by the staff of the Cardiopulmonary Laboratory, Kelsey Seybold Clinic, National Aeronautics and Space Administration/Johnson Space Center, Houston, Texas. The subjects of this study were 277 males who elected to have a graded exercise stress test as part of their annual health examination. Data describing the subjects at rest and exercise are provided in Table. All tests were administered on a treadmill (Quinton Model 18-60) utiliz- ing the Bruce protocol. The stress test was terminated when the subject either reached voluntary exhaustion or was terminated early by the physician in charge. During the stress test, blood pressure was simultaneously measured by BPMS and a trained technician at: (1) rest, (2) each minute of exercise, and (3) each minute during an 8-min recovery period. The procedures followed by the technician were: (1) the BPMS inflated the cuff to the target pressure; (2) using a stethoscope, the technician listened for Korotkoff sounds; and (3) at the appearance and disappearance of the sounds, read the systolic and diastolic (phase V) pressures directly from the mercury manometer. The BPMS provided pressures which were automatically stored in the memory of an LS 11 microcomputer. The technician was "blind" to the blood pressure readings of BPMS. The data analysis sought to compare the mean differences of blood pressures measured simultaneously by the two methods, and intraclass analysis was used to evaluate the consistency of exercise blood pressure measured by each method within each stage of the Bruce protocol. The data were analyzed by two work intensities: first, percentage of age predicted heart rate and second, minutes of the test. Results Provided in Table are the descriptive statistics of the subjects studied. The subjects varied considerably in terms of age and physical working capacity. All blood pressure values in Table were technician-determined. The resting values were taken after lying quietly for five minutes. The exercise values were recorded during the last minute of work. TABLE Characteristics of patients studied Variable Mean Standard deviation Range Physicial characteristics Age (yr) 42.9 Height (cm) Weight (kg) 78.5 Resting characteristics Heart rate (beats/min) 63.5 Diastolic blood pressure " 74.8 Systolic blood pressure " Peak exercise characteristics Treadmill time (min) 9.6 Heart rate (beats/min) Diastolic blood pressure " 81.3 Systolic blood pressure " a Blood pressures are auscultatory with resting pressures taken in a supine position and exercise values recorded at the final minute of exercise.

3 J. A. Garcia-Gregory et al.,' Automated blood pressure measurement 317 The resting and recovery blood pressure readings were simultaneously measured by technician and B})MS. At rest; the mean systolic pressures were not statistically different, with a difference of only 1 mmhg. The mean resting diastolic pressure of BPMS was significantly (p<0.01) lower than the pressure obtained by technician, but the difference was small, only 3 mmhg At recovery, the mean systolic reading of BPMS was significantly (p <0.01) higher (6 mmhg) than the technician reading and the recovery diastolic means were nearly identical to the differences found at rest. Furnished in Fig. 1 is exercise blood pressure response as a function of treadmill time and percentage of agedetermined maximum heart rate (HRmax). These data are consistent with expected exercise blood pressure variation: diastolic pressures remained relatively stable and systolic pressures steadily rose with work intensity. For all stages of work, the systolic readings of BPMS were significantly (p<o.o) higher than technician values. As work intensity increased, the differences in systolic pressure became more pronounced. At peak workloads, the difference in systolic pressure was about 20 mmhg. For all stages of work the diastolic pressures of BPMS were significantly (p<o.o) lower than technician readings. The difference range was 5-10 mmhg. The intraclass analysis is furnished in Fig. 2. This analysis was designed to examine the degree to which blood pressure measured by each method could be expected to vary due to random fluctuations within a 3-min workstage. The analysis strategy was to use each of the three blood pressure readings within a workstage for the analysis. Provided are the standard errors of measurement for a blood pressure reading for each workstage. For all levels of work, the standard errors were lower for technician-determined values than for BPMS. At peak workloads, the standard errors for exercise systolic pressure of BPMS were over three times higher than the technician-determined values. The standard error is an estimate of the random variation, and these data showed that BPMS was subject to more random fluctuations than a trained technician. The final analysis conducted was to examine the correlations of exercise blood pressure measured by both methods for various stages of work, and correlations were computed for the middle minute of each stage of work. The coefficients for exercise blood pressure measured by BPMS and technician were for systolic pressures and for diastolic pressures. n general, the lowest correlations were found at the more intensive workloads when a patient was more likely to be jogging rather than walking. Discussion The results of this study showed that exercise blood pressure measured with BPMS varied systematically from readings obtained by trained technicians. As the workload increased, the differences became more pronounced. Unexpectedly, the technician-determined readings were more consistent than the automated readings. Glasser and Ramsey (1981) found that systolic pressures obtained with an automated system were higher than values obtained by trained technicians. The differences in exercise systolic readings ranged from 5 to 12 mmhg over the workloads studied. These systematic differences led Glasser and Ramsey to conclude that at a systolic pressure of 200 mmhg, the automated system could be expected to read 9-15 mmhg higher than simultaneously recorded manual readings. The average exercise diastolic differences between automated and manual measured blood pressure reported by Glasser and Ramsey were slight; the differences were less than 3 mmhg, which are smaller than the values we found. However, the differences were in the same direction; lower diastolic pressures were obtained with the automated system. t is suspected that the smaller diastolic differences found by Glasser and Ramsey can be traced to equipment. For both studies, phase V Korotkoff sounds were used by the technician. The BPMS used by NASA followed a set criterion for registering diastolic pressure, whereas the system evaluated by Glasser and Ramsey recorded phase V if the pressure was greater than 80 mmhg, but phase V if the pressure was less than 80 mmhg. A major problem with exercise diastolic pressure is that Korotkoff sounds can sometimes persist at very low' pressures. The BPMS tended to register some low values which would account for the lower diastolic pressures recorded by the NASA system. One group of investigators has reported that it is difficult to obtain accurate exercise diastolic readings (rving et al. 1977). Automated systems are believed to be more sensitive in determining the onset of Korotkoff sounds than the human ear (Glasser and Ramsey, 1981). This would be one explanation for the higher systolic readings recorded by automated systems. A second reason, however, is that if a noise artifact was present, it would tend to "trigger" the automated reading above the "true" systolic pressure. This possibility can be seen by examining a hardcopy tracing of blood pressure obtained by BPMS (Fig. 3). The top reading clearly delineates the systolic and diastolic pressures, whereas the Korotkoff sound patterns of the bottom reading are not clearly apparent. We suspect that some of the recorded sounds are exercise artifacts which would tend to produce higher systolic and lower diastolic pressures. With the rapid advances in microcomputers and electronic technology, an attractive option is to automate procedures in an effort to reduce human error. This is one suggested advantage of automated blood pressure systems. The intraclass analysis (Fig. 2) showed that the technician-determined readings were, surprisingly, more consistent than the automated readings. An important consideration for these results is that the technician was

4 318 Clio. Cardiol. Vol. 7, May 1984! ::: A C ,'l Treadmill Time (min) ! ::: 160!! t:-- ff,. A &- 60 1: < >85 B % Maximum Heart Rate FG. Exercise systolic and diastolic blood pressure trends for given levels of work. Plots are by (A) minutes of Bruce protocol and (B) percentage of HRmax. Both show systematic differences at each stage of work. (6-6, Technician-determined readings;.-., BPMS readings.)

5 1. A. Garcia-Gregol)' et al.: Automated blood pressure measurement 319 Systolic pressure Diastolic Pressure 15 5 o A Treadmill Time (min) Systolic Pressure Diastolic Pressure 15 5 o -L- < >84 < >84 B % Maximum Heart Rate FG. 2 Standard emus of measurement for exercise blood pressure measured by technician (0) and BPMS (.) for (A) Bruce protocol and (B) HRmax. The figures show that BPMS was subject to more random variance than technician-measured blood pressure.

6 320 Clin. Cardio!. Vol. 7, May 1984 r-r-r-, r "T' r " ' / ;!...l : --. i-',.. ; " ; - c-. -' -1-,-,--1-, Tt! i - :.,,-' t- J i.. i i L -t-:.,,,:'-- 1 '(-! + '1- -., L J. -L ' =!:-v-=f "o-..."=-t"...:..., -=t-"---.,;"=' """"'/""'"'"=r-"-=! T-R' i i " f/f)t i. i 'i... : ll' 'b,' P! i"--, '" r'\ '" '" '" j f! j._., t+ r. n. b - f-- T, j! '-,.. r-..4 L flo t. b t. ft, ft. "'- 1l ft,. '" "'...' - Fro. 3 Hardcopy of blood pressure tracings from BPMS. Top tracing shows the clear location of the Korotkoff sounds and the lower tracing isolates " noise" at the extremes of the blood pressure ramp. ft.. highly trained and experienced. The intraclass analysis used the three blood pressure readings within each 3-min workstage with the assumption that each patient was internally consistent over this constant workload. The standard error of measurement is a statistic that represents the degree to which blood pressure could be expected to vary owing to chance. These analyses showed that at each workstage, the technician was more consistent than BPMS. We suspect this may be partly due to extreme scores caused by arti facts recorded by the BPMS. The technician used in this study was consistent; a reading was not recorded until it was certain a Korotkoff sound was heard. f there was doubt, the pressure was immediately retaken by the technician. These data showed that the technician was more consistent than the BPMS, but the intraclass analysis does not show that the technician was more accurate. Steinfeld et ai. (1981) compared resting blood pressures measured by electronic and stethoscopic sphygmomanometers with intra-arterial pressures. The correlations between auscultatory and intra-arterial pressures were somewhat higher than the coefficients found between intra-arterial and electronic pressures. Bergman et al. (1972) reported that the correlations between the intra-arterial exerci se blood pressure and the readings obtained by BPMS and auscultatory methods were similar. Exercise blood pressure was measured while riding a bicycle ergometer, and these readings were less subject to the noise introduced by walking or jogging on a treadmill. The results of these studies do not support the conclusion that automated methods are more accurate than manual readings while walking or jogging on a treadmill. n our experience, a trained technician can be a safeguard for screening obviously erroneous readings which would be recorded by automated methods. Glasser and Ramsey (l981) reported correlations of 0.96 and 0.85, respectively, between concurrent manual and automated systolic and diastolic pressures. These correlations are higher than the coefficients we found, but the findings are not directly comparable. Glasser and Ramsey calcul ated a si ngle coefficient for systolic and diastolic pressure. These high correlations were computed on the patient's blood pressure measured at rest, several stages of work, and recovery. By using all stages of work, recovery, and rest, the correlations were inflated because more variability was introduced into the bivariate relationship. This was especially true for systolic pre sure which changes su bstantially with exercise. The coeffi cients reported for our data represent the relation between BPMS and manual determinations for each defined stage of work. The results of this study

7 J. A. Garcia-Gregory et al.: Automated blood pressure measurement 321 showed that the correlation between simultaneously measured BPMS and technician-detennined blood pressure was lower at a defined workload and that the correlations were somewhat lower at more intense workloads when the subject was jogging. There is a growing trend to define exercise hypertension for men at 230 mmhg systolic (Sannerstedt, 1981) and an exercise hypotensive response :s 140 mmhg. systolic (Bruce et al., 1977; rving et al., 1977), and exercise hypertensive responses have been shown (Jackson et al., 1983; Wilson and Meyer, 1981) to be predictive of the development of resting hypertension. These criteria were developed with nonnative data obtained by the auscultatory method. Using the technician readings, only 2.9% of the NASA employees were found to be hypertensive during the exercise test and none were hypotensive. The difference between systolic pressure measured by BPMS and technician at maximum work levels tended to be about 20 mmhg. f BPMS was used to measure exercise blood pressure, about 15% of the NASA executives would be judged to be hypertensive during exercise. The data presented in Fig. 1 shows the difference is present at the lower levels of work. f the concern of exercise blood pressure response was a hypotensive response because of left ventricular disfunction and a criterion of a systolic pressure of :s 140 mmhg was used, the automated system would identify few hypotensive patients. f automated systems are to be used, the data need to be compared with technician readings so that meaningful interpretations can be made. The results of this study showed that exercise blood pressures measured by an automated system were substantially different than pressures measured by a trained technician. The differences become more pronounced with exercise intensity, and the automated pressure readings were more prone to random fluctuations than readings by trained technicians. References Bergman SA, Blomquist G, Mitchell J, Hoffler GS: Skylab blood pressure measuring system: Correlative study with intm-arterial pressures. Unpublished study, Southwestern Medical School, Dallas (1972) Bruce RA, DeRouen RA, Peterson D, rving JB, Chinn N, Blake B, Haffey, V: Noninvasive predictors of sudden cardiac death in men with coronary heart diesase: Predictive value of maximal stress testing. Am J Cardiol 39, 833 (1977) Glasser SP, Ramsey MR, : An automated system for blood pressure detennination during exercise. Circulation 63, 348 (1981) rving JB, Bruce RA, DeRouen TA: Variations in and significance of systolic pressure during maximal exercise (treadmill) testing. Am J Cardiol 39, 841 (1977) Jackson AW, Squires WG, Grimes G, Beard EF: Prediction of future resting hypertension from exercise blood pressure. J Cardiac Rehab 3, 263 (1983) Nolte RW: Automated blood pressure measuring system (M092). n Biomedical Results from Skylab. RS Johnson and LF Dietlein (Eds.). NASA, Washington, D.C. (1977) Sannerstedt R: Exercise in the patient with arterial hypertension. Prac Cardiol 7, 89 (1981) Steinfeld L, Cole A, Nazarian H, Richard A: A comparison of electronic and stethoscopic sphygomomanometers. Circulation (Abstr.) 64, (No.4) V-213 (1981) Wilson NV, Meyer BM: Early prediction of hypertension using exercise blood pressure. Prevent Med 10, 62 (1981)