Key words: Motion-analysis, referee, football, METs, oxygen consumption.

Subjects
The procedures adopted in the present study are in agreement with the 96/1996 resolution from the Conselho Nacional de Saúde (National Council of Health), which deals with research procedures in human beings. The project was approved by the Ethic committee of the University Hospital of the Federal University of Paraná (UFPR). All the subjects were volunteers male referees accredited by CBF (Brazilian Confederation of Soccer) that were informed verbally and by a written form about the nature and demands of the study, as well as about eventual health risks. They were also informed that they could withdraw from the study at any time. Written consent was obtained from each individual after approval of the experimental protocol by the Ethic Committee, including an authorization to film their activities during the match. The participants were submitted to the official physical tests used by FIFA to evaluate their referees and to a medical evaluation prior to this study. All subjects received approved in both evaluations. These referees officiated in 29 matches of the 2005 and 2006 Paraná Championship of professional Soccer, Series A and B. The State of Paraná (Brazil) Football Federation responsibly by the organization of this championship consent to realize this study, including film referees activity during the matches and made all the other procedures.

Each match was video-recorded using a camera (Sony, model Handycam CCD-TRV 128) attached to a tripod positioned at the side of the pitch, at the halfway line, at a height of about 15 m and at a distance of about 15-20m from the field. The camera filmed the referee close up to evaluate locomotive activities. The following locomotive categories were used: standing (0 km·h^-1), walking (5.83 km·h^-1), jogging (8.85 km·h^-1), running (11.37 km·h^-1), sprinting (18.28 km·h^-1), and backwards running (8.85 km·h^-1). These categories were chosen in accordance with a previous study (Da Silva and Rodrigues-Añez, 1999), whereas the mean speed for each category was determined after studies of the videotapes. The time for the subject to pass knows distances pre-markers in the field was used to calculate the speed for each locomotive activity. The frequency and duration of each activity were digitally recorded by the same experienced observer. These data were utilized to calculate the distance covered by the referee in each activity. The total distance covered during each stage of the match was calculated by adding the distances covered in each motion activity. This time motion analysis methodology utilized in the present study is similar to that employed by other authors (Catterall, 1993; Krustrup, 2001). Reproducibility of the results obtained in the present study was determined in a pilot study. Five matches were analyzed twice by the same observer, separated by at least 3 months. No systematic differences were observed in the test-retest analysis and the intra-individual differences in total distance covered were less than 0.25km. These results of reproducibility and reliability were similar to that obtained by the same observed in a previous study (Da Silva and Rodrigues-Añez, 1999).

Energy expenditure values were estimated from the time they spent in each motor activity. To calculate the oxygen uptake during running and sprinting we utilized the equation: VO₂ = 3.5 + (0.2 x m·min^-1), and for walking the equation: VO₂ = 3.5 + (0.1 x m·min^-1), both suggested by the American College of Sport Medicine (1988). Oxygen uptake during jogging and backwards running was calculated by the equation suggested by Bubb et al., 1985, VO₂ = 3.5 + (km·h^-2 x 0.394). Energy expenditure during the time subjects remained still was calculated multiplying the basal metabolism constant 3.5 ml·kg^-1·min^-1 by the mean body weight. The value of consumption of O₂ was then transformed in kcal, multiplying it by 5 (Wilmore and Costill, 2005).

The determination of referee's energy expenditure during the match in METs was possible by measuring the VO₂ spent for the execution of each locomotor activity. For an adult, 1 MET is approximately 3.5 ml of consumed O₂, for kilogram of body mass per minute (1 MET = 3.5 ml.kg^-1.min^-1). Therefore, the energy expenditure was estimated by dividing the VO₂ of each motion action by the constant 3.5.

In the present study, it was observed that during officials matches soccer referees perform physical aerobic activities of low and moderate intensity, and present a significant decline in energy expenditure between the first and second half. Other studies involving soccer referees have also confirmed that the physical activity of referees during match-play predominantly involves the utilization of energy originating from aerobic metabolism (Asami et al., 1988; Castagna et al., 2004; Catterall et al., 1993; Da Silva and Rodriguez-Añez, 1999; Johnston and McNaughton, 1994; Krustrup and Bangsbo, 2001).

During the match, referees covered an average distance of 9155.4 ± 70.3 meters (8411 - 9765), a value similar to that observed in other studies (Catterall et al., 1993; Johnston and McNaughton, 1994; Krustrup and Bangsbo, 2001), but lower than that observed by Castagna et al., 2004, in Italian referees. The differences observed in match coverage between these studies could be partially related to different competitive level of the official matches analyzed. The referee's total distance covered during the match is similar to that described for soccer player, in particular to midfield players (Bangsbo et al., 1991; Mohr et al., 2003; Rienzi et al., 2000; Tumilty, 1993). The total distance covered provides an overall index of work rate, based on the assumption that the energy expenditure during the match is directly related to total work output (Drust et al., 1998). Also, the distance covered at high intensity activities (running and sprinting) is correlated with referee's aerobic fitness (Castagna, 2002; Krustrup, 2001).

During a soccer match the referee must observe the actions of players in an area that measures 8.250 m² on average (Castagna et al., 2007). Approximately every 4-6 seconds, the referee changes motion activity (D'Ottavio and Castagna, 2001; Krustrup and Bangsbo, 2001), equating to 1268 different activities during 90 minutes of an official match. Of these, 588 are consequence of low-intensity activities (standing, walking, jogging) and 161 of high intensity (running and sprint) (Krustrup and Bangsbo, 2001). These indicate that soccer referring is a highly intermittent exercise mode. In other words, soccer referees have significant aerobic energy expenditure throughout a match and episodes of considerable anaerobic energy turnover. The aerobic energy production accounted for approximately 90% of total energy consumption (Bangsbo, 1994). This percentage present a large inter-individual differences due to the variety of factors which influence exercise intensity. All the evidence presented above support the necessity of improving aerobic metabolism in soccer referees, but without forgets anaerobic fitness (Castagna et al., 2007).

In the present study, a similar pattern was observed with referees spending 52% of the total match time walking, 19% jogging and 15% standing. These prevalence of a pattern of movements at low speed (walking and jogging) have been observed in others studies (Asami et al., 1988; Catterall et al., 1993; Da Silva and Rodriguez-Añez, 1999; D'Ottavio and Castagna, 2001; Johnston and McNaughton, 1994; Krustrup and Bangsbo, 2001), which a range between 41.8% and 73.8% of match coverage (Castagna et al., 2007). The results of this study were found to be in the upper limit of this range (71%). It has previously been described that low-intensity activities are negatively correlated with aerobic fitness (Castagna and D`Ottavio, 2001). Differences in game style and match intensity among countries are other factors that should also be taken into account when results are compared (Rebelo et al., 2002; Helsen and Bultynck, 2004). For example, it has been reported that South American international soccer players cover significantly less total distance during a match (8638meters) than English premier league players (10104meters) or elite Danish players (10800meters) (Rienzi et al., 2000; Bangsbo et al., 1991). Recently, it has been reported that work-rates of referees during match-play were partly related to the physical activities of the players (Weston et al., 2007). The high percentage (71%) of movements at low speed (walking and jogging) found in the present study could be the result of a lower match intensity of the regional championship where referees officiated, and/or differences in game style between south American and Europeans players. A recent work with English referees found that match intensity (evaluated by ratings of perceived exertion - RPE) was correlated to competition standard (Weston et al., 2006). Other studies observing match intensities, player's physical activities and referee work-rates in the same match of official Brazilian soccer championship are necessary to explain this point.

Running at high (sprint) and moderate intensity speeds (running) are rarely performed by soccer referees. In the present study this did not occur for more than 6% of total time (Table 1). They covered a mean distance of 1010.9 ± 74.5 meters running and 122.7 ± 19.3 meters sprinting, without a significant difference in performance between the two halves (Table 2 and Figure 1). In other studies, it has been reported that the percentage of match time spent performing high intensity activities varies from 4 to 18% (Asami et al., 1988; Catterall et al., 1993; Da Silva and Rodriguez-Añez, 1999; Johnston and McNaughton, 1994; Krustrup and Bangsbo, 2001; Rebelo et al., 2002), with a significant reduction during the second half (Krustrup and Bangsbo, 2001) a result suggesting that referees experience fatigue towards the end of the match. In the present study, a significant reduction was not observed for these activities during the second half, however, a significant reduction in backwards running was observed (see Figure 1). This type of motor activity represented 6% of total match time. We know from several studies that backwards running is more demanding in terms of energy expenditure than forward running (Flynn et al., 1994; Reilly and Bowen, 1984; Williford et al., 1998). The results of this study could therefore suggest that referees experience fatigue towards the end of the match. However, the absence of a significant difference in high intensity activities between halves does not allow us to confirm this point. Factors such as depletion of glycogen stores, dehydration and hyperthermia may contribute to the development of fatigue in the later stages of a soccer game. These factors were better studies in soccer players than in referees (Bangsbo et al., 2006; Castagna et al., 2007; Da Silva and Fernandez, 2003; Reilly, 1997). The most important aspect of refereeing is the decision-making process, but the effect of physical performance on this aspect has received little attention. Helsen and Bultynck, 2004 quantified the perceptual cognitive workload of elite soccer referees and observed that they undertakes about 137 observable decisions during a match, uniformly distributed throughout the six 15min intervals. On the other hand, mean distance from infringements increased in the second period of the match only in the left attacking zone, a zone where the ability of assistant referees to help was limited (Mallo et al., 2007). This observation reinforces the idea of fatigue at the end of the match, and its possible negative effect on the decision-making process. At the present time is not clear if improving exercise capacity may decrease the probability of incurring judgments errors during the match.

Energy expenditure during a physical activity can be measured directly in laboratory or estimated indirectly from oxygen uptake or from core temperature. Several reports have shown that it is possible to estimate the energy expenditure of a physical activity with reasonable precision using the consumption of oxygen (Hopkins and Powers, 1982; Coast and Welch, 1985; Daniels, 1985; Daniels and Daniels, 1992). Indirect measurements are generally used to estimate energy expenditure during soccer matches (Bangsbo et al., 2006). As described above, measures of heart-rate were utilized in several studies to estimate exercise intensity and aerobic involvements of soccer players and referees (Bangsbo et al., 2006; Castagna, 2007). Factors like dehydration, hyperthermia, and mental stress elevate the heart rate without affecting oxygen uptake, leading to an overestimation of this variable. In the present study the VO₂ was estimated for each motion action. A principal limitation of the present time-motion technique may be the reliability of the match observer. No systematic differences were observed in the test-retest analysis and the intra-individual differences in total distance covered were less than 0.25km. Further studies are necessary to measure directly VO₂ uptake during the match, through online measurements, and compare it against the VO₂ values estimated from time-motion analysis or recording of heart rate.

In the present study, a soccer referee's estimated energy expenditure during the match was 734.7 ± 11.9 kcal (903.2 - 641.3). This energy expenditure was significantly reduced in the second half (see Figure 2), a fact that could be explained by the reduction in jogging and back- wards activities observed in this period. Direct measurements of oxygen consumption in soccer players during a match reported a value of energy expenditure of 1195kcal, and the value estimated by the recording of heart rate was 1565kcal, in mean (Shephard, 1992). These values of energy expenditure are significantly lower than the value of 1702 kcal reported in English referees by Weston and Brewer (2002), who determined in laboratory conditions, a HR-VO₂ relationship to estimate this parameter. The differences observed between field and laboratory assessment of energy expenditure in referees is not completely clear, but it is known that the treadmill protocol of determining HR-VO₂ could overestimate the match VO₂. As described above backward movement elicits a greater metabolic demand and cardiopulmonary response than forward locomotion (Flynn et al., 1994; Reilly and Bowen, 1984; Williford et al., 1998). Williford et al., 1998, working with collegiate tennis players, found that at running speed (8 km·h^-1) the VO₂ and heart rate were both 15% greater for backwards running compared with forward running. Furthermore, the energy expended per minute by soccer players at 9 km·h^-1 at backwards movements increased 4.93kcal over forward running, an increase of approximately 40% (Reilly and Bowen, 1984). The equations utilized in the present study to calculate the oxygen uptake during motor activities use the mean speed for each category. Backwards displacements happen at the same mean speed that jogging (8.85 km·h^-1), so the calculated value for oxygen uptake is the same. This is a limitation to our study, because the expected greater metabolic demand with backward movements is probably underestimated by the equation. If we utilized the data of Williford work to correct the VO₂ in 15%, the estimated total energy expenditure during the match will be 768,1kcal, a value still lower than that reported by Weston and Brewer (2002).

If we take into consideration the fact that a male person consumes, on average, 2.900 kcal·day^-1 (National Research Council, 1996), then the referee's energy demand during a match would be from 3.500 to 4.000 kcal/day. A professional soccer player, in order to cope with the daily physical activity and the physical effort due to a soccer training and competition, should consume 3500 to 4.500 kcal·day^-1 (Clark, 1994; Bangsbo et al., 2006; Ebine et al., 2002; Rico-Sanz, 1998; Shephard, 1992). A soccer player's training is generally more frequent and longer in duration than a referee, with a higher energy cost (Bangsbo et al., 2006). Considering that the energy expenditure of referees is lower to that for a player during a match, their energy needs to support daily physical activities (training and refereeing) will generally be lower.

Based on the total or partial caloric expenditure of each action expressed in kcal, we could not determine the intensity of the physical activity. However, with the estimated VO₂ of each motion action it is possible establish another way to enunciate the energy expenditure and also of classifying the physical activity according to its intensity, using MET (Metabolic Equivalent). As discussed above referees spent 52% of the match walking, an activity with an estimated consumption of oxygen of 13.25 ml·kg^-1·min^-1, which corresponds to an energy expenditure of 3.8 METs. Another action of low consumption of oxygen during the match is the time spent standing, with an estimated consumption of oxygen of 3.5 ml·kg^-1·min^-1, and an energy expenditure of 1 MET. Therefore, during 67% of the match, they perform physical activities that could be classified as low intensity, aerobics (Ainsworth et al., 1993; Durnin and Passmore, 1967). The remaining 32% of match time is spent performing physical activities using energy originating from the aerobic and anaerobic system (backwards running, jogging, running and sprint). In backwards running and jogging, the estimated consumption of O₂ was 34.52 ml·kg^-1·min^{- 1}, which corresponds to 9.8 METs, an activity considered as intense (Durnin and Passmore, 1967). During only 7% of match time referees performed running or sprinting movements, with an energy consumption of 41.46 VO₂ ml·kg^-1·min^-1 (11.8 METs) and 64.46 VO₂ ml·kg^-1·min^-1 (18.4 METs). Activities with energy consumption above 10 METs are classified as extremely intense (Durnin and Passmore, 1967).

To obtain an unbiased measure of the mean intensity of referee's physical activity during the match, the mediated average instead of the simple arithmetic average was used. The amount of METs expenditure depends on the speed, divided in 6 categories in this study. For example, the amount of METs spent in the condition "standing still" is relatively lower than during sprint. Besides, there is a great difference in the total time of permanence in activities of high and low speed. Therefore, measurement of the mediated average significantly reduces the possibility of obtain biased results by the highlighted reasons. Using the mediated average, it was estimated that the mean energy consumption of a referee is equal to 5 METs. Based on this result and using the classification proposed by Ainsworth et al., 1993, it is possible to classify the referee's physical activity as moderate energy intensity. This result coincides with the data obtained in the studies that measured the intensity of referee's physical activity by means of heart rate (Catterall et al., 1993; Castagna, 2001; Da Silva and Rodriguez-Añez, 2005; Helsen and Bultynck, 2004; Johnston and McNaughton, 1994; Krustrup and Bangsbo, 2001; Rebelo et al., 2002).

The METs system was developed to standardize intensities of physical activities and promote comparison across different studies (Ainsworth et al., 1993). The METs values obtained in the present study, allow us to compare the intensity of a referee's physical activity with the results obtained in soccer players or athletes of other sports, which energy requirements were better establish. This information could be useful in the future to establish a specific nutritional program for soccer referees.