Journal of Applied Physiology

Validity of the heart rate deflection point as a predictor of lactate threshold during running

John A. Vachon, David R. Bassett Jr., Stephen Clarke


During an incremental run test, some researchers consistently observe a heart rate (HR) deflection at higher speeds, but others do not. The present study was designed to investigate whether differences in test protocols could explain the discrepancy. Additionally, we sought to determine whether the HR deflection point accurately predicts lactate threshold (LT). Eight trained runners performed four tests each:1) a treadmill test for maximal O2 uptake,2) a Conconi test on a 400-m track with speeds increasing ∼0.5 km/h every 200 m,3) a continuous treadmill run with speeds increasing 0.5 km/h every minute, and4) a continuous LT treadmill test in which 3-min stages were used. All subjects demonstrated an HR deflection on the track, but only one-half of the subjects showed an HR deflection on the treadmill. On the track the shortening of stages with increasing speeds contributed to a loss of linearity in the speed-HR relationship. Additionally, the HR deflection point overestimated the LT when a continuous treadmill LT protocol was used. In conclusion, the HR deflection point was not an accurate predictor of LT in the present study.

  • Conconi test
  • exercise
  • endurance
  • performance
  • blood lactate

lactate threshold (LT) is an important variable in the field of physiology, because it closely predicts actual performance in endurance events such as distance running (2, 11, 17, 36). Conconi et al. (9) proposed a method for noninvasively determining LT in runners. Their results show the expected linear relationship between heart rate (HR) and running speed at submaximal speeds but a plateau in HR at high running speeds. These investigators report that the deflection point of the HR-running speed relationship occurs at the same speed as the LT (9). They report that this method is applicable to other endurance sports, including cycling, racewalking, rowing, skating, and swimming (4, 5, 8, 10, 15).

Conconi’s method (9) is controversial for two reasons. First, many physiologists report that there is a linear HR relationship during incremental, maximal exercise tests or that HR only reaches a plateau in a certain percentage of subjects (3, 20, 29, 31, 33). However, Conconi et al. (9) reported that HR leveled off in all 210 of the runners they initially tested and in all 65 of the runners tested in a subsequent study (4). A second problem is that numerous authors have reported that the HR deflection point overestimates the directly measured LT (23, 24, 27, 33), contrary to observations of Conconi and co-workers (4, 6-10, 15). Recently, Conconi et al. (10) acknowledged the controversy by listing a large number of studies that refute as well as support their hypotheses.

The present study was designed to assess the validity of Conconi’s method, which uses HR measurements to predict LT. The first aim of the study was to determine whether HR is linearly related to running speed or whether it reaches a plateau at higher running speeds (9). The second aim was to determine whether Conconi’s method results in an HR deflection point that accurately predicts LT during a continuous treadmill test.



Eight male distance runners, training 30–60 miles/wk, volunteered to participate in the study. Subjects were instructed not to perform hard physical training during the 48 h before each test. The clothing, shoes, and environmental conditions, as well as all equipment used, were consistent for each subject and were recorded to establish controlled experimental conditions. The ambient temperature and humidity were carefully monitored, since environmental conditions can influence HR and blood lactate levels (1, 18). Air temperatures for the track, treadmill, and LT tests were 21.8 ± 6.2, 21.3 ± 1.7, and 20.8 ± 2.5°C, respectively. Relative humidity values for track, treadmill, and LT tests were 57.4 ± 12.6, 49.3 ± 12.5, and 53.1 ± 9.4%, respectively. The sequence of tests was counterbalanced to minimize the influence of an order effect.

The nature of the study was explained to the subjects, and they were given an opportunity to ask questions about anything that was unclear. All subjects signed an informed consent form approved by the university’s Institutional Review Board. Anthropometric measurements were taken on each subject, including height, weight, and sum of skinfolds. The physical characteristics of the subjects are shown in Table 1.

View this table:
Table 1.

Physical characteristics of the subjects

Maximal exercise test.

After a stretching and warm-up period, a graded maximal treadmill test was performed to determine the subject’s maximal O2 uptake (V˙o 2 max). Subjects ran at a comfortable speed (12–16 km/h) on the level for 1 min, and then the grade was increased by 1% at 1-min intervals for as long as the subject could continue. A two-way breathing valve with a mouthpiece was used, and a noseclip was worn to prevent nasal breathing. Ventilation was measured at 1-min intervals with a calibrated airflow meter (RAM 9200, Rayfield Equipment, Waitsville, VT). Percentages of expired gases were measured with O2 and CO2 analyzers (Applied Electrochemistry, Sunnyvale, CA) connected to a computerized system. These analyzers were calibrated against known gases analyzed by the micro-Scholander technique (30). O2 uptake (V˙o 2) and CO2 production were calculated using standard metabolic equations (28). Attainment ofV˙o 2 max was based on achieving two of the following three criteria:1) a plateau inV˙o 2 (<50% of the predicted increase inV˙o 2) with increasing speeds,2) respiratory exchange ratio >1.10, or 3) HR within 10 beats/min of age-predicted maximal HR (220 − age). HR was measured using a heart watch (Vantage XL, Polar Electro, Kempele, Finland).

Conconi’s protocol.

Subjects underwent a 10 min warm-up at 50% HR reserve. They then performed a continuous run of 8–12 laps around a 400-m track, starting at a speed of 10–14 km/h and finishing at a final speed of 18–22 km/h. The subjects were instructed to increase running speed ∼0.5 km/h every 200 m until they could no longer continue. The investigators called out projected 200-m split times to the runners, and the actual running speeds were later computed from the 200-m split times stored in the stopwatch. HR was recorded continuously at 5-s intervals with use of the Polar heart watch; these data were later downloaded to a computer for analysis.

Laboratory treadmill protocol.

Subjects underwent a 10-min warm-up at 50% HR reserve. They then completed 16–20 stages on a treadmill (model 2Y-72, Quinton, Seattle, WA), beginning at an initial speed of 11–12 km/h and finishing at 18–21 km/h. The stage duration was held constant at 1 min. The running speed was increased by ∼0.5 km/h for as long as the subject could continue. During each stage of the test, the number of treadmill drum revolutions was recorded with a mechanical counter; these data were later used to establish the precise running speed. An electric fan in front of the subject was used to provide convective cooling during the test. HR values were recorded as previously described.

LT test.

Subjects performed treadmill running of 8–10 stages, beginning at a speed corresponding to ∼60% of theirV˙o 2 max. Speed was increased by ∼0.8 km/h in each stage. During the last 30 s of each 3-min stage, a 1.0-ml venous blood sample was obtained. At the end of each 3-min stage, the treadmill speed was increased to the next predetermined speed. This routine was continued until all stages were completed.

Blood lactate measurement.

Blood samples were taken from a 20-gauge Teflon catheter (Critikon, Tampa, FL) inserted into a forearm vein. A three-way stopcock was inserted into the catheter, and patency of the line was maintained with a heparin-saline solution (110 USP U/ml). Before a blood sample was obtained, the catheter and the stopcock were flushed by withdrawing a volume of fluid equal to twice the dead space of the catheter into a syringe. Then a 1-ml blood sample was drawn into another syringe, and 100–200 μl were immediately placed in a tube containing an antiglycolytic (sodium fluoride) and a cell-lysing agent (cetrimonium bromide), and the tube was shaken. The lactate concentration of the blood sample was analyzed within 45 min.

Blood lactate concentrations were determined using an automated lactate analyzer (2300 STAT plus, Yellow Springs Instruments, Yellow Springs, OH). The analyzer was calibrated with known lactate standards (5.0 and 15.0 mM).

Data analysis.

The HR vs. running speed data for the Conconi track and indoor treadmill protocols were fitted using a third-order polynomial equation (Cricket Graph III, Islandia, NY). The third-order polynomial equation was chosen, because it provided a significantly better curve fit than a linear or second-order polynomial (r 2 = 0.983 for track, r 2 = 0.991 for treadmill). For the Conconi protocol, the HR deflection was independently determined by two investigators using visual inspection. The blood lactate data were fitted using an exponential equation. LT was defined as the running speed corresponding to the break point where blood lactate concentrations increased from baseline (34). In other words, the threshold was considered to be the highest speed that could be attained before an increase in the blood lactate level was seen. By use of this definition, the LT was determined by visual inspection. A Student’s paired t-test was used to compare the speeds corresponding to the directly measured LT and the HR deflection point. The significance level was set atP < 0.05.


In all subjects, the track protocol of Conconi et al. (9) resulted in an HR deflection at high running speeds. However, four of the eight subjects showed no signs of an HR deflection on the treadmill; subsequently, they will be referred to as group 1 (Fig. 1). The other four subjects demonstrated an HR deflection on the track and the treadmill tests; they will be referred to as group 2 (Fig. 2).

Fig. 1.

Blood lactate and heart rate (HR) responses to 3 continuous, incremental run tests. Subjects 1, 2, 3, and 4(A, B, C, andD, respectively) showed an HR plateau on Conconi track protocol, in which 200-m stages were used, but not on a treadmill protocol, in which 60-s stages were used. HR deflection point (top arrow) occurred at higher speeds than lactate threshold (LT; bottom arrow).

Fig. 2.

Blood lactate and HR responses to 3 continuous, incremental run tests.Subjects 5, 6, 7, and8 (A, B, C, and D, respectively) showed an HR plateau on Conconi track protocol and treadmill protocol, in which 60-s stages were used. HR deflection point (top arrow) occurred at higher speeds than LT (bottom arrow).

Figures 3 and4 show the time course of the HR response in a typical runner from group 1(subject 2) for the continuous treadmill run and the Conconi test, respectively. The maximal HR achieved on the Conconi test and the treadmill test was 180 beats/min. The maximal running speed achieved was higher on the Conconi test (18.3 km/h) than on the treadmill (17.3 km/h).

Fig. 3.

Time course of HR response for subject 2 during each stage of a continuous treadmill run test (60-s stages). Final HR achieved at each stage increased by a fairly constant amount, indicating a linear speed-HR relationship.

Fig. 4.

Time course of HR response for subject 2 during each stage of a continuous Conconi test (200-m stages). HR increased by a constant amount, except for last stage, where despite a large increase in speed (1.2 km/h), HR increased by only 1 beat/min. This resulted in a leveling off in speed-HR relationship.

Figures 5 and6 show the time course of the HR response in a typical runner from group 2(subject 5). The maximal HR achieved on the Conconi test was similar to that achieved on the treadmill test (186 and 185 beats/min, respectively). The maximal speed achieved was higher on the Conconi test (21.2 km/h) than on the treadmill (19.9 km/h).

Fig. 5.

Time course of HR response for subject 5 during each stage of a continuous treadmill run test (60-s stages). During last 3 stages, runner was unable to raise his HR above 185 beats/min. This subject showed an HR plateau on treadmill and Conconi protocol.

Fig. 6.

Time course of HR response for subject 5 during each stage of a continuous Conconi test (200-m stages). Runner attained much faster speeds on Conconi test than on treadmill. However, his maximal HR still could not go much higher than 185 beats/min.

LT in the subjects occurred at running speeds of 12.72–16.47 km/h. The HR deflection point occurred at 16.01–20.37 km/h (Table2). LT and HR deflection data forsubject 4 were not considered in the statistical analysis, because precise determination of LT was compromised by lack of a stable baseline. For the remaining subjects the difference in speeds represented by HR inflection and LT was statistically significant (P< 0.0008). The relationship between the speed at LT and HR deflection speed (r = 0.688) is depicted in Fig. 7.

View this table:
Table 2.

Speeds corresponding to LT on a continuous LT test and HR deflection on a Conconi-style outdoor track test

Fig. 7.

LT speed vs. HR deflection speed for 7 subjects. Dashed line, line of best fit; solid line, line of identity.


One reason for the controversy surrounding Conconi’s method (9) is that other researchers do not always find an HR deflection during an incremental exercise test (20, 29, 31, 33). One of the main features of Conconi’s method is that as the test proceeds, the amount of time required to complete each stage decreases. With Conconi’s track protocol, which results in a shortening of stage durations, all the runners demonstrated an HR deflection.

Using a treadmill protocol with constant 60-s stages, only one-half of the runners showed a noticeable HR deflection. These findings are consistent with the work of Astrand and Rodahl (3), Ekblom et al. (16), and Davies (13). These researchers noted a strong, linear relationship between work intensity and cardiac frequency at submaximal workloads. However, at near-maximal efforts, they found that some individuals demonstrated a slight deviation from linearity as HR began to plateau. In more recent studies, in which constant stage durations of 45–60 s were used, an HR deflection occurs in 45–71% of subjects (22, 23, 29). Thus it would appear that only about one-half of all individuals show an HR deflection with constant stage protocols, whereas virtually all subjects show an HR deflection with Conconi’s protocol (4, 5, 8-10, 15, 21). The unique feature of the present study is that it compared test protocols with constant and those with shortening stages in the same runners.

It has been hypothesized that the shortening of stages with Conconi’s protocol may result in insufficient time for attainment of steady-state HR at high speeds. If this were true, it would explain the loss of linearity in the speed-HR relationship. However, Conconi et al. (9) argue against this hypothesis, noting that HR adapts to each new speed within 10–20 s. Most of our runners were able to achieve a steady-state HR (±2 beats/min) on Conconi’s protocol within 15–30 s (although subject 3required 45 s). Thus we agree that Conconi’s protocol normally allows adequate time for HR to adapt to each new speed (Figs. 3-6).

Then why do some runners show an HR deflection on Conconi’s protocol but a linear HR response on the treadmill? We believe that Conconi’s test, because of the shortening stages, decreases the lactic acid accumulation, thus reducing muscle fatigue. This allows the runner to continue increasing his speed beyond the stage where his HR has reached its maximum level. Most of the runners in group 1 were able to achieve maximal running speeds on the track that were 1.0–2.5 km/h faster than on the treadmill, even though their maximal HR had already been attained. Thus the runners ingroup 1 showed an HR plateau on Conconi’s protocol but a linear speed-HR relationship on the treadmill (Fig. 1).

Because of concerns about the shortening stages, in 1996 Conconi et al. (10) changed the test protocol so that HR measurements were based on fixed time intervals rather than on fixed distances. Every 30 s the runner increased his speed by a constant amount (until the last part of the test, when he accelerated over several stages). Using their 1996 constant-stage protocol, they found that an HR deflection occurred in 95% of their subjects and 99% of athletes who regularly perform the test. Thus they observe that an HR deflection occurs in almost all individuals, even when the stage duration is held fairly constant. This conflicts with our study, where we found that a “constant-stage” protocol yielded a linear HR response in one-half of our subjects. It is likely that the shorter stages (30 s) and rapid acceleration phase in Conconi’s new protocol allow subjects to attain higher maximal speeds than a test with constant 60-s stages. Thus, similar to their earlier protocol, the 1996 protocol appears to increase the likelihood of observing an HR deflection.

The second major finding in this study was that the HR deflection point (with Conconi’s 1984 protocol) occurs at much higher speeds than the directly measured LT. In addition, there was only a modest correlation between the speed at LT and HR deflection speed (r = 0.688). These observations agree with the findings of several other investigators during graded cycle ergometry (20, 23, 27) and running (24, 33). In contrast, Conconi et al. (9) showed extremely close agreement between LT and HR deflection (r = 0.99). The controversy surrounding this issue is difficult to resolve. However, we believe that the most likely explanation is that it is due to differences in the LT test protocol.

Traditionally, LT is determined by one of two methods:1) a continuous test with 1-, 3-, or 4-min stages (23, 26, 29, 32, 34, 35, 37) or2) a discontinuous test where a series of three sets of 10-min runs (interspersed with 15 min of rest) are performed over several days (2, 12, 17, 19, 35). We used a continuous test with 3-min stages, which has been proven to yield LT values similar to a discontinuous test with 10-min stages (35). However, Conconi et al. (9) employed an unconventional LT test consisting of 1,200-m stages with 15 min of active recovery between stages. With Conconi’s LT test, a runner would complete 1,200 m in ∼4–6 min compared with the usual 10-min discontinuous stage. Because Conconi’s stages are about one-half as long as a typical discontinuous stage, this may decrease the blood lactate accumulation, causing an artificial “shift” of the LT curve. This appears to be the reason that they report close agreement between HR deflection point and LT (7-10, 15, 25), whereas researchers using more conventional LT tests have found that the HR deflection overpredicts LT (23, 24, 27,33).

We considered the effects of air resistance during the physiological testing. Our LT tests were conducted on a treadmill, whereas those of Conconi et al. (9) were conducted on an outdoor track. During treadmill running we used a fan to provide convective cooling; this created a head wind of 2.8 km/h. This meant that during the track test the runners faced a relative head wind that was 7–16 km/h greater than it would have been during the treadmill LT test. This would have increased the gross energy cost by only ∼1–3% (14). Thus differences in air resistance between treadmill and track running have very little effect and cannot explain the 19% difference we observed between LT and the HR deflection point.

In the present study, “LT” was defined as the running speed corresponding to the break point where the blood lactate concentration increases from baseline (34). Conconi and co-workers use the term “anaerobic threshold” to refer to this same physiological phenomenon (4, 7, 9). Other common definitions for LT include a fixed blood lactate concentration of 2.5 mM or a 1.0 mM increase in blood lactate from the baseline levels (34).

In summary, Conconi’s protocol (9) is more likely to result in an HR deflection than conventional treadmill protocols. This is partly due to the shorter stage durations in Conconi’s protocol, which may decrease lactic acid accumulation, thus reducing muscle fatigue. This allows a runner to continue increasing his running speed beyond the point where maximal HR is attained. In addition, the HR deflection point on a Conconi test occurs at a higher running speed than the measured LT. The reason that some researchers find good agreement between these two variables (4, 7-10, 15, 25), whereas others do not (23, 24, 27,33), may relate to differences in the LT test protocol.


The authors thank Dixie L. Thompson, Renan M. Sampedro, and Edward T. Howley for assistance in preparing the manuscript.


  • Address for reprint requests and other correspondence: D. R. Bassett, Jr., The University of Tennessee, Knoxville, Exercise Science Unit, 1914 Andy Holt Ave., Knoxville, TN 37996-2700.

  • J. A. Vachon was a scholar in the University of Tennessee-Knoxville Division of Biology Threshold Program funded by the Howard Hughes Medical Institute. Support was also provided by the University of Tennessee, Knoxville, Exhibit, Performance, and Publication Expense Fund.

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.


View Abstract