Heart rate and shot timing in archery: Carrillo 2011

Seventeen male archers, seven novices and ten experienced, were monitored beat by beat during an indoor competition at 18 metres. The clearest difference between the groups is not cardiac but temporal: the experienced archers took 28.1 seconds per arrow on average against 20.2 for the novices, and shot time turned out to be the strongest predictor of the score of a single arrow.

For heart rate variability, computed over the 20 seconds before each release, the experienced archers had higher RMSSD, higher pNN50 and a higher low frequency band, while at rest, before the competition, the two groups were not statistically distinguishable. Comparing each archer with his own resting values says more: during the competition the experienced archers lost 18.4 normalised units of low frequency activity on average and gained 10.2 points of pNN50, while in the novices the same indices barely moved (-4.0 and -2.6). Within the end the two groups went in opposite directions, with high frequency activity rising from the first to the third arrow in the experienced archers and falling in the novices.

One result complicates the usual reading. In the experienced archers the rise in high frequency activity between the first and the second arrow came together with a drop in score, and the scores of the second and third arrows were lower than that of the first. The final model, holding shot time, low frequency activity and the ratio between the bands, accounts for about 12% of the variation in score.

The competition was an indoor event in Greece run under FITA rules, split into two rounds separated by a 30 minute break: in each round, ten ends of three arrows in two minutes at 18 metres, with two minutes between ends, for twenty ends and sixty arrows in all. Before the start, heart rate variability was recorded for five minutes with the athlete seated, and that value served as the baseline.

Inter-beat intervals were collected with a chest-strap heart rate monitor (Polar RS800CX) and analysed with dedicated software. For each arrow, shot time was timed by hand and defined as the interval between the archer reaching into the quiver for the arrow and the release; the last 20 seconds of that time were used to compute the indices. In the time domain, RMSSD, the square root of the mean of squared differences between successive beats, and pNN50, the percentage of intervals differing from the previous one by more than 50 milliseconds. In the frequency domain, the low and high frequency bands in normalised units and their ratio. The score of every arrow was recorded.

The analysis crosses group, position of the arrow within the end and end number in multivariate analyses of variance, one on absolute values and one on changes from baseline, and closes with a stepwise multiple linear regression looking for predictors of the arrow score.

Seventeen boys, all male, one competition, 18 metres indoors. Experience goes together with age (a mean of 14.0 years in the novices against 16.5 in the experienced archers), so what separates the groups may be maturation rather than training. The age standard deviations printed in the paper, ±8.5 and ±10.3 years, are not compatible with two groups of adolescents, and the sample cannot be reconstructed any better than that.

The frequency indices are computed over 20 second windows. For the low frequency band that is a short window, since those oscillations have periods of several seconds, and the authors themselves accept that the higher low frequency activity in the experienced archers may reflect breath control rather than autonomic modulation. Parasympathetic activity is not measured: it is inferred from RMSSD and pNN50. The signal comes from a chest strap, not from an electrocardiogram.

The degrees of freedom in the regression, 996, show that the thousand arrows are treated as independent observations, with no account taken of the fact that they come from 17 archers shooting sixty times each, so the standard errors come out smaller than they should be. The effect sizes are nearly all small, between 0.005 and 0.1 of partial eta squared, the exception being shot time and score (0.24). Absolute scores are not reported, only that the experienced archers shot better: how far apart the two groups were on the target is not known.

What an archer can take from this is the finding about time, not the one about the heart: in this competition, what was most closely tied to the score of a single arrow was how long passed between picking up the arrow and the release, and the experienced archers took eight seconds longer. It remains a description of two different groups, not a demonstration that taking longer raises the score, because an archer with four years behind them has everything else as well.

On the cardiac side the study offers no lever. The indices separate the groups, but they account for a small slice of the variation in scores, and within the end the sign runs the other way for anyone who reads parasympathetic activity as shorthand for a good shot: in the experienced archers the high frequency index rose from the first to the third arrow while the score fell. A heart rate strap, here, is a tool for observing, not a criterion for deciding when to let go.