Gross efficiency (GE) [the ratio of mechanical power output (PO) to metabolic power input (PI) (van Ingen Schenau and Cavanagh
1990)] is an important factor in performance (Joyner
1991; Joyner and Coyle
2008) and in the use of energy flow models (de Koning et al.
1999,
2005). Hettinga et al. (
2007) showed that a change in GE of only 0.9% could result in a 25.6 s difference in time over a 20 km cycling time trial. Additionally, differences in GE (or its equivalent, running economy) have been shown to account for differences in cycling (Lucia et al.
2002) and running (Foster and Lucia
2007; Ingham et al.
2008) performance, particularly in athletes matched for high
VO
2max. However, for the proper use of energy flow models and for the interpretation of research findings it is important to have good insight into the magnitude and variation in GE.
Different studies have shown that there are differences in GE between subjects, which could be due to differences in technique or skill (Nickleberry and Brooks
1996; Moseley and Jeukendrup
2001; Hintzy et al.
2005; Hopker et al.
2007) and/or to genetics (Coyle et al.
1991,
1992; Mogensen et al.
2006) of the subjects. Most studies performed on efficiency are cross-sectional in nature and longitudinal studies are needed to study the causal relationship between, for example training and GE. Prior to conducting a longitudinal study it is essential to have good insight into the reliability of measuring GE. Moseley and Jeukendrup (
2001) determined economy (EC), delta efficiency (DE), and GE on three different occasions, separated by at least 5 days. The within subjects coefficient of variation (CV) was 3.3, 6.7, and 4.2% for EC, DE, and GE, respectively. Thus, when measuring efficiency on multiple days a smaller variation in GE is expected than in DE. The mean CV was 3.2% for GE, which implies that a change in GE as small as 0.6% (e.g., 20.0–20.6%) can be perceived (Moseley and Jeukendrup
2001). However, there are some limitations to the study of Moseley and Jeukendrup (
2001). The first limitation is that Moseley and Jeukendrup (
2001) averaged the oxygen uptake (
VO
2) and RER data over the second and third minute of each 3-min exercise step to determine GE. From previous research it is known that
VO
2 needs 3 min to reach a steady-state and it is therefore not ideal to average
VO
2 and RER data over the second and third minute because,
VO
2 may not be in steady-state (Whipp and Wasserman
1972; Barstow and Molé
1991). The second point of discussion is that Moseley and Jeukendrup (
2001) calculated GE as the mean of all breath-by-breath data collected in the last 2 min of each exercise intensity step, during which RER did not exceed 1.0. It is well known that GE increases in a curvilinear fashion with an increase in exercise intensity (Ransom et al.
2008; Ettema and Lorås
2009), so the most accurate way to determine GE is to calculate GE from the breath-by-breath data collected at the highest exercise intensity with an RER <1.0. Therefore, the purpose of this study was to determine the reliability of GE using some improvements in the research design of Moseley and Jeukendrup (
2001).
Besides the variation in GE between days the within day variation in GE also needs to be considered. Circadian (or diurnal) rhythms in resting heart rate (HR) (Faria and Drummond
1982; Giacomoni et al.
1999; Callard et al.
2001), oral temperature (Souissi et al.
2004), mesenteric temperature (Callard et al.
2000), rectal temperature (Deschenes et al.
1998; Giacomoni et al.
1999), blood pressure (Deschenes et al.
1998), and circulating hormones (Deschenes et al.
1998) have been extensively investigated. Body temperature (rectal temperature) shows a circadian rhythm in rest, with a mean amplitude of 0.44°C and a mean acrophase at 17:16 h, which persisted at light, moderate and heavy exercise (Reilly and Brooks
1990). However, there are conflicting results about the circadian effect on aerobic exercise capacity. Hill (
1996) and Giacomoni et al. (
1999) found significant differences in submaximal steady-state
VO
2 between exercise performed in the morning and in the evening. The aerobic system responded faster (Hill
1996) and reached a greater
VO
2 amplitude in the evening (Hill
1996; Giacomoni et al.
1999). This time of day effect on the cardiovascular and respiratory response to exercise can be partly attributed to the circadian rhythm in body temperature (Hill
1996). Subjects performed incremental exercise tests at 08:00, 12:00, 16:00, and 20:00 h randomized over different days, with at least 48 h between consecutive tests in the study of Deschenes et al. (
1998). They found no significant effect of time of day on either pre-exercise or exercise
VO
2. These inconclusive results could be due to the chosen time points at which the exercise tests were conducted. In order to determine the circadian rhythms in
VO
2 and other exercise related variables ideally, exercise bouts should be evenly distributed over 24 h (Nelson et al.
1979; Souissi et al.
2004). If a circadian rhythm in submaximal
VO
2 is present, as reported by Hill (
1996) and Giacomoni et al. (
1999), it would be expected that GE will also vary with time of day. Brisswalter et al. (
2007) investigated the effect of time of day on net efficiency (NE) and GE. Subjects performed four submaximal exercise bouts, two in the morning (between 07:00 and 08:30 h) and two in the evening (between 19:00 and 20:30 h), at 80% of the PO associated with the ventilatory threshold (80% PT
vent). Exercise bouts were separated by at least 24 h. No significant time of day effect was found in
VO
2,
VCO
2, or RER at rest or during light cycling exercise (45 W). Nevertheless, when exercise intensity increased to 80% PT
vent there was a significantly higher
VO
2 amplitude and a larger
VO
2 time constant (slower response) in the morning compared to the evening, resulting in a significantly higher NE in the evening (17.3 vs. 20.5%). The difference in GE, 15.1% (morning) versus 17.1% (evening), did not reach statistical significance, which agrees in principal with Moseley and Jeukendrup (
2001) who found a mean CV in GE of only 3.2%. The main shortcoming of the study of Brisswalter et al. (
2007) is that they did not use the minimum of six exercise bouts evenly distributed over the 24 h of the day, which is viewed as a critical issue by Nelson et al. (
1979). A study that did use six exercise bouts evenly distributed over 24 h was performed by Reilly and Brooks (
1990), who found a significant circadian rhythm at rest in rectal temperature, HR,
VO
2,
VCO
2 and VE. However, during submaximal exercise no circadian rhythm in
VO
2,
VCO
2, VE, NE, and GE was found. The limitation of the study of Reilly and Brooks (
1990) is that NE and GE were determined during exercise at 82 and 147 W, which are relatively low absolute workloads, corresponding to relative workloads of 37 and 56%
VO
2max. To accurately investigate a possible circadian rhythm in efficiency each subject should exercise at the same relative intensity and multiple exercise intensities should be chosen, in order to be sure that the highest efficiency is reached (Ransom et al.
2008). Because previous research showed that muscle temperature affected efficiency during in vitro measurements (He et al.
2000) and during cycling exercise (Ferguson et al.
2002; Bell and Ferguson
2009) it could be expected that the circadian rhythm in body temperature affects GE during cycling exercise.
Therefore, the purpose of this study was to assess the variation in GE between and within days by measuring GE at six time points equally distributed over the 24 h of the day. When the variation in GE within and between days is known it will be easier to interpret the findings of other studies and to accurately use energy flow models.