High-Altitude Football
On the 28th May, 2007, football's governing body, FIFA, banned international matches from being
played at more than 2,500m (8,200ft) above sea level [BBC News].
FIFA said the decision was made because of concerns over players' health and possible distortion of competition.
This ruling will primarily affect countries in South America which is home to over half of the World Cup winning teams.
The continent's natural environment provides the ideal training ground; extreme differences in altitude present serious challenges
to teams playing away games at higher altitudes. International football games take place in high altitude cities
such as Bogotá, Colombia (2,600 m), Quito, Ecuador (2,800 m), and La Paz, Bolivia (3,600 m).
Players have to cope with adverse physiological conditions, such as breathlessness, fatigue,
and possibly acute mountain sickness [1-6].
Physical activity can exacerbate these symptoms, preventing players from performing to
their full capacity [7]. Ability to consume oxygen, which is reduced by acute exposure to altitude,
reflects players’ physical fitness and correlates with football performance at a national level [8].
Although it is recognized that acclimatized high altitude teams suffer less than their lowland opponents,
the relationship with football performance has not been demonstrated before.
Here we show that the home-away altitude difference significantly affects the outcome of international football games. Our analysis of South American games shows that the challenge of acclimatisation and subsequent deterioration in physical fitness have a significant impact on the outcome of the match. High altitude teams have an advantage over their sea level opponents when playing at high altitude but also suffer a disadvantage when playing away at sea level.
We obtained scores of international football games played in South
America between 1900 and 2004 in order to directly assess the influence of altitude on
football. Only home and away games were included; all matches played in neutral
venues were omitted. This dataset contained football scores for ten national teams
and a total of 1,460 games.
Figure 1 compares the percentage of wins at home and away for each country in South America. Although this information
provides an approximate indication of the skill of each team, it fails to account for the variation due to the physiological effects of altitude.
The top two teams (Brazil and Argentina) are from seal level and also perform extremely well when playing at home.
McSharry (2008) accounted for the differing levels of skill by normalising the number of goals scored and conceded for each game relative
to the mean value of goals scored and goals conceded for each country separately. He then investigated the ability of a range of parametric nonlinear models (linear through to cubic) to
describe the normalised data where the altitude difference between the home and away teams was included as an explanatory variable. Model selection was performed using
the Bayesian information criterion which suggests that the quadratic model is most parsimonious in terms of balancing the goodness of fit with complexity.
The results (Figure 2) reveal that when the two teams were from the same altitude
(altitude difference equals zero) the probability of the home team winning was 0.43, but this rose to 0.49 if a theoretical team from ~3600m (such as Bolivia) played
at home against a theoretical opponent from sea level. However, the probability of a home win rose to an even greater extent, to 0.67, for a theoretical team from sea
level (such as Brazil) playing at home against a team from ~3600m (such as Bolivia). This analysis demonstrates that sea level teams going up to altitudes of
~1000-2000m are not compromised in their ability to win; in fact at 1000m the probability of the home team beating their sea level opponent falls marginally to 0.42.
Conversely, a theoretical team from ~2000m altitude competing against a sea-level opponent at sea level has a much reduced likelihood of a win compared with playing at home.
Overall, this analysis reveals that teams from moderate to high altitude are much less likely to win away than at home. Theoretically, the approach
controls for the historical performance of teams, any influence of skill and differences between managers (McSharry, 2008), and reflects just the effects of altitude.
This statistical analysis suggests a reduced ability to win at sea level for football
teams that are residents of moderate/high altitude, which is consistent with anecdotal reports of lassitude and general malaise upon descent (personal communication, Bengt Saltin).
Further information is available from Patrick McSharry and in the associated British Medical Journal article, also see this BBC News report (in Spanish). Suggestions and comments are welcome.
[1] T.D. Brutsaert, H. Spielvogel, R. Soria, M. Araoz, E. Caceres, G. Buzenet, V. Mercedes, M. Paz-Zamora, E. Vargas (2000), Performance of altitude acclimatized and non acclimatized professional football (soccer) players at 3,600 m. Journal of Exercise Physiology, 3(2): pp. 28-37.
[2] C.C. Monge (1932), Acclimatization in the Andes. Johns Hopkins Press, Baltimore, MD.
[3] M. Ward, J.S. Milledge, J.B. West, M.P. Ward (2000), High Altitude Medicine and Physiology, 3rd ed., Arnold, London.
[4] J.R. Sutton, G. Coates, C.S. Houston (Eds). (1992) The Lake Louise consensus on the definition and quantification of altitude illness. Hypoxia and Mountain Medicine. Queen City Printers, Burlington, Vermont.
[5] A.J. Peacock (1998), Oxygen at high altitude. BMJ, 317:1063-1066.
[6] P.W. Barry, A.J. Pollard (2003), Altitude illness. BMJ, 326:915-919.
[7] C.M. Maresh, M.R. Deschenes, R.L. Seip, L.E. Armestron, K.L. Robertson, B.J. Noble (1993), Perceived exertion during hypobaric hypoxia in low- and moderate-altitude natives. Med. Sci. Sports Exerc., 25, 945-951.
[8] U. Wisloff, J. Helgerud. J. Hoff (1998), Strength and endurance of elite soccer players. Med Sci Sports Exerc. 30:462–7.