Journal Logo

Special Communications: Methods

Validity of predicting mean arterial blood pressure during exercise


Author Information
Medicine & Science in Sports & Exercise: December 1999 - Volume 31 - Issue 12 - p 1876
  • Free


During exercise blood flow to skeletal muscle is tightly coupled to its metabolic rate. This is achieved by increases in cardiac output and mean arterial blood pressure (MAP) and regional redistribution of blood flow through adjustments in vascular resistance. MAP is thus the major determinant of tissue perfusion and a key variable in many basic cardiovascular equations. MAP is defined as the average pressure exerted by the blood on the arterial wall during a complete cardiac cycle or series of cardiac cycles (3) and is thus affected by its site of measurement, the configuration of the pressure wave form, and the duration of diastole. In the resting state, MAP is conventionally estimated as being the product of one-third the systolic pressure plus two-thirds the diastolic pressure (or diastolic pressure plus one-third pulse pressure) (1,10). Since the configuration of the pulse wave form is known to change with sympathetic stimulation and the duration of diastole to decrease (as heart rate increases) during exercise, one might question the validity of this prediction equation during exercise of varying intensities. To our knowledge this has never been tested by invasive measurement during the exercise state. Our purpose was thus to record directly MAP over a wide range of exercise intensities and to examine the accuracy of the standard prediction method for estimating it from measurements of systolic and diastolic pressure.


The data presented in this paper were collected from three separate studies. In one (Study 1), subjects performed progressive cycle ergometry to fatigue, while in the others they performed continuous constant load exercise at either 65% V̇O2max (Study 2) or at 100 W (Study 3).


Eight healthy male volunteers (23 ± 4 yr, weight 79 ± 6 kg) participated in Study 1. All were subjects for a larger investigation (7) and had experienced brachial arterial catheterization on 3–5 occasions before collection of data for the present study. Subjects in Study 2 were 13 young males (24 ± 2 yr, weight 74 ± 8 kg) who were also participating in a larger study (5) and had experienced catheterization on one or two previous occasions. Subjects for Study 3 were six males and two females (23 ± 4 yr, weight 79 ± 4 kg) participating in an ongoing study. Subjects for all studies were informed of the potential risks associated with the procedures and provided written consent. All procedures were approved by the Human Research Ethics Committee of McMaster University.


In Study 1 subjects cycled on an electrically-braked ergometer at 0 W (unloaded pedaling at ∼70 rpm), 60, 120, 180, 240, and 300 W. They exercised at each power output until a steady-state heart rate (HR) was achieved. This was considered to occur when HR ceased to increase and variations were within ± 2 beats·min−1. On average this occurred within 2–2.5 min at each stage. Once steady state was achieved, subjects continued to cycle for an additional 60 s, over which beat-to-beat blood pressure and the integrated MAP were recorded from the brachial artery.

In Study 2 power outputs were calculated for each subject, which would elicit an oxygen consumption equivalent to 65% V̇O2max as established on a previous progressive test. They maintained this power output for 15 min, with blood pressure being recorded directly from the brachial or radial arteries.

In Study 3, all subjects cycled for 10 min at a power output of 100 W, and blood pressure was recorded from the radial artery.


In Study 1 blood pressure was recorded from an indwelling pressure-tip transducer (Millar, Mikro-Tip, Houston, TX) introduced into the brachial artery and advanced to midsternal level. In Studies 2 and 3 blood pressure was recorded by coupling a 1.5-inch 20-gauge catheter in the radial (N = 19) or brachial arteries (N = 2) to an external Novatrans pressure transducer (model MX807, Medex, Hilliard, OH) as we have previously described (4). For both systems, pressure signals were amplified by an Acuda amplification system (model 143, Windaq/200 DataQ Instruments Inc., Akron, OH), sampling at a frequency of 300 Hz. The linearity of both systems was verified over a range of 0–500 mm Hg by calibration against a strain gauge reference transducer. Calibration was also done before and after each test with a mercury manometer by injecting pressures between 0 and 200 mm Hg. We have previously cross-referenced the pressure-tip transducer system against the fluid-column system by catheterization of both arms and found the values and wave formations to be identical (6).

Average systolic and diastolic pressures and MAP (area under the curve) were measured over 30-s windows. MAP was then also estimated over this time as the product of the average diastolic pressure plus one-third of the average pulse pressure. Since there were relatively large inter-individual differences in HR at each power output, it was decided to normalize blood pressure measurements to each individual’s heart rate. This was achieved by post-hoc analysis of blood pressure data over the 30-s duration in which HR was 100, 110, 120, and so on up to 200 beats·min−1. To ensure that blood pressure was recorded during steady-state conditions (i.e., at a given power output), it was necessary to enlarge the window by ± 4 beats·min−1. Thus, the predicted MAP was calculated from the average systolic and diastolic pressures at each HR (± 4 beats·min−1) and the true MAP (integrated area under the curve) directly measured over the same 30 s. In Study 1 all subjects were able to maintain exercise until a HR of 190 beats·min−1, but only five were able to achieve a HR of 200 beats·min−1.

MAP was also measured and estimated in the resting state in the above 29 subjects and in an additional 20 subjects who were participants in other studies. These data have been pooled and added to the present paper for comparison purposes.


Actual and estimated values for MAP are summarized in Figure 1 for the resting state and for a range of heart rates from 100 to 200 beats·min−1. Regression lines for these data are presented in Figure 2 and indicate a largely linear increase in actual (r = 0.97) and predicted (r = 0.88) MAP for a given increase in HR.

Figure 1
Figure 1:
Actual (•) and predicted (○) MAP at rest and at different exercise heart rates from 100 to 200 beats·min−1. Values are means and SE. N = the number of subjects for each plot.
Figure 2
Figure 2:
Linear regression lines for actual (solid line) and predicted (dotted line) mean arterial blood pressure data presented inFig. 1.

On average, predicted values tended to be within ± 4–5 mm Hg of actual values. The greatest discrepancies occurred at heart rates of 110 beats·min−1, where predicted values overestimated actual values by 9 mm Hg; 140 beats·min−1, where predicted values overestimated actual values by 10 mm Hg; and 200 beats·min−1, where predicted values underestimated actual values by 10 mm Hg. In the resting state, estimated MAP was within 3 mm Hg of actual MAP.


We interpret our data as indicating that the predictability of the conventional method for estimating MAP (diastolic pressure plus one-third of pulse pressure) remains minimally affected by exercise throughout a range where HR more than doubles. Thus, the relationship between the area under the pulse pressure curve during diastole and the magnitude of the pulse pressure is, for the most part, preserved as HR increases (diastole decreases). In effect, the loss in area resulting from changes in the configuration of the pulse wave and the decrease in diastole is compensated for by the increase in pulse pressure that occurs with an increase in systolic pressure (with little change in diastolic pressure). This is illustrated in Figure 3 where the pressure traces are presented for the same subject at a HR of 64 beats·min−1 (rest) and at a HR of 140 beats·min−1 (100 W). Blood pressures averaged 143/73 and 175/73 mm Hg, respectively. Actual MAP was 96 mm Hg at rest and 106 mm Hg during exercise, with estimated MAP being 96 and 107 mm Hg, respectively.

Figure 3
Figure 3:
Radial arterial blood pressure traces for the same subject at rest (dotted line) and while exercising at 100 W on a cycle ergometer (solid line).

It should be noted that in this study estimates of MAP were based on direct measurements of systolic and diastolic pressures. Because resting and steady-state exercise blood pressures are continuously oscillating, calculations of MAP from auscultatory estimates of blood pressure would be considerably less accurate. These oscillations (as illustrated in Fig. 4) are primarily second-order or high-frequency oscillations caused by ventilatory dynamics (both tidal volume and tidal frequency) and third-order low-frequency oscillations resulting from variations in sympathetic efferent activity (8). Such oscillations in pressure may be as great as 25 mm Hg from peak to trough, even though HR is constant and the subject is in a steady-state exercise condition. It is thus apparent that auscultatory estimates of blood pressure are prone to considerable sampling error, confounded by the fact that the diastolic pressure which is registered represents a different cardiac cycle than that in which the systolic pressure was registered. When one combines this with the inherent limitations of the auscultatory method (auditory acuity, choice of which Korotkoff sound to denote diastolic pressure, rate of cuff deflation relative to HR, etc.), it is apparent that during exercise the reliability of the technique is even more questionable (for further discussion, see references (2) and (9)).

Figure 4
Figure 4:
Sample trace from the brachial artery of a subject performing steady-state exercise on a cycle ergometer at 120 W. Note oscillations in systolic and diastolic blood pressure.


Estimating MAP as the product of diastolic pressure plus one-third of the pulse pressure results in values that are quite close to actual mean pressure. The accuracy of this prediction equation does not change during exercise over a wide range of heart rates. Thus, in instances where blood pressure is estimated by auscultatory methods, failure to accurately predict mean pressure is the result of faulty estimates of systolic and/or diastolic pressures and not limitations in the equation per se.

The data presented in this paper were collected from studies funded by the Natural Sciences and Engineering Research Council of Canada and the Defense and Civil Institute of Environmental Medicine, contract W7711–9-7091/01-XSE.


1. American College of Sports Medicine. Resource Manual for Guidelines for Exercise Testing and Prescription. Philadelphia: Lea & Febiger, 1988, p. 51.
2. American College of Sports Medicine. Guidelines for Exercise Testing and Prescription, 5th Ed. Philadelphia: Lea & Febiger, 1995, p. 96.
3. Guyton, A. C. Textbook of Medical Physiology, 3rd Ed. Philadelphia: W. B. Saunders, 1966, p. 353.
4. Haslam, D. R., N. McCartney, R. S. McKelvie, and J. D. MacDougall. Direct measurement of arterial blood pressure during formal weightlifting in cardiac patients. J. Cardiac Rehab. 8:213–225, 1988.
5. MacDonald, J. R., J. D. MacDougall, M. A. Tarnopolsky, S. A. Interisano, K. Smith, and E. V. Younglai. Hypotension following mild bouts of resistance exercise and submaximal dynamic exercise. Eur. J. Appl. Physiol. 79:148–154, 1999.
6. MacDougall, J. D., R. S. McKelvie, D. E. Moroz, D. G. Sale, N. McCartney, and F. Buick. Factors affecting blood pressure during heavy weightlifting and static contractions. J. Appl. Physiol. 73:1590–1597, 1992.
7. MacDougall, J. D., R. S. McKelvie, D. E. Moroz, J. S. Moroz, and F. Buick. The effects of variations in the anti-G straining maneuver on blood pressure at +Gz acceleration. Aviat. Space Environ. Med. 64:126–131, 1993.
8. Pagani, M., D. Lucini, O. Rimoldi, et al. Low and high frequency components of blood pressure variability. Ann. N.Y. Acad. Sci. 783:10–23, 1996.
9. Rasmussen, P. H., B. A. Staats, D. J. Driscoll, K. C. Beck, H. W. Bonekat, and W. D. Wilcox. Direct and indirect blood pressure measurement during exercise. Chest 87:743–748, 1985.
10. Vander, A. J., J. H. Sherman, and D. S. Luciano. Human Physiology: The Mechanisms of Body Function, 3rd Ed. Toronto: McGraw-Hill, 1980, p. 283.


© 1999 Lippincott Williams & Wilkins, Inc.