Eur J Appl Physiol (2010) 108:599–606
DOI 10.1007/s00421-009-1253-9
ORIGINAL ARTICLE
Influence of cold water face immersion on post-exercise
parasympathetic reactivation
Hani Al Haddad • Paul B. Laursen •
Said Ahmaidi • Martin Buchheit
Accepted: 8 October 2009 / Published online: 31 October 2009
Ó Springer-Verlag 2009
Abstract The aim of the present study was to investi-
gate the effect of cold water face immersion on post-
exercise parasympathetic reactivation, inferred from heart
rate (HR) recovery (HRR) and HR variability (HRV)
indices. Thirteen men performed, on two different occa-
sions, an intermittent exercise (i.e., an all-out 30-s Win-
gate test followed by a 5-min run at 45% of the speed
reached at the end of the 30–15 Intermittent Fitness test,
interspersed with 5 min of seated recovery), randomly
followed by 5 min of passive (seated) recovery with
either cold water face immersion (CWFI) or control
(CON). HR was recorded beat-to-beat and vagal-related
HRV indices (i.e., natural logarithm of the high-frequency
band, LnHF, and natural logarithm of the square root of
the mean sum of squared differences between adjacent
normal R–R intervals, Ln rMSSD) and HRR (e.g., heart
beats recovered in the first minute after exercise cessa-
tion) were calculated for both recovery conditions. Para-
sympathetic reactivation was faster for the CWFI
condition, as indicated by higher LnHF (P = 0.004), Ln
rMSSD (P = 0.026) and HRR (P = 0.002) values for the
CWFI compared with the CON condition. Cold water
face immersion appears to be a simple and efficient
Communicated by Dag Linnarsson.
H. Al Haddad (&)
S. Ahmaidi
M. Buchheit
Research Laboratory, EA-3300 Laboratory of Exercise
Physiology and Rehabilitation, Faculty of Sport Sciences,
University of Picardie, Jules Verne, 80025 Amiens, France
e-mail: alhaddad.hani@gmail.com
P. B. Laursen
School of Exercise, Biomedical and Health Sciences,
Edith Cowan University, Joondalup, WA, Australia
means of immediately accelerating post-exercise para-
sympathetic reactivation.
Keywords Short-term recovery

Autonomic nervous system
Vagal-related indices

Face immersion
Introduction
Physical exercise causes an increase in sympathetic activ-
ity, concomitant with parasympathetic withdrawal, result-
ing in higher heart rates (HR). Conversely, post-exercise
cardiodeceleration is mediated by a progressive increase in
parasympathetic activity (Imai et al. 1994), as well as a
continued sympathetic withdrawal (Savin et al. 1982).
Postponement of parasympathetic reactivation after exer-
cise has been shown to be associated with an increased risk
of sudden cardiac death (Billman 2002). Indeed, para-
sympathetic activity is thought to afford a cardioprotective
background. Thus, sympathetic hyperactivity (Billman
2006) or reduced vagal tone (Smith et al. 2005) following
exercise may confer a poor cardioprotective background
inducing ventricular fibrillation and sudden cardiac death.
For example, a lowered heart rate recovery (HRR), indic-
ative of a reduced parasympathetic activity, is frequently
reported in chronic heart failure cases (Cole et al. 1999;
Imai et al. 1994). Similarly, a decrease in post-exercise
vagal-related HR variability (HRV) indices, which provide
insight into the level of ‘parasympathetic reactivation’
(Buchheit et al. 2007a; Goldberger et al. 2006), has
been observed in patients with coronary artery disease
(Goldberger et al. 2001).
Means of improving post-exercise parasympathetic
reactivation are receiving greater interest. For example,
123
600
cold (Buchheit et al. 2009b), as well as thermoneutral
(Miyamoto et al. 2006) water immersions, have been shown
to be efficient methods of triggering parasympathetic
activity and accelerating parasympathetic reactivation.
While hydrostatic pressure likely plays a major role in
triggering parasympathetic reactivation (i.e., by reducing
vascular capacitance and shifting peripheral blood volume
into the compliant thoracic vasculature leading to an increase
in central blood volume), stimulation of cold receptors has
been hypothesized to be another important parameter that
might have a cumulated effect on parasympathetic nerve
activity (Buchheit and Laursen 2009; Buchheit et al. 2009b;
Mourot et al. 2008). However, implementing water baths in
the field setting can be problematic.
Selective immersion of the face in cold water (namely
cold water face immersion, CWFI) has been reported to
be a very simple and non-invasive means of inducing
bradycardia (i.e., diving reflex) at rest (Finley et al. 1979;
Hayashi et al. 1997; Paulev et al. 1990) and during
exercice (Finley et al. 1979; Smith et al. 1997). For
example, Hayashi et al. (1997) showed that CWFI-
induced bradycardia without breath-holding increased
vagal-related HRV indices at rest, suggesting that face
immersion itself, and not presumably the apnea, was the
likely trigger for the increased vagal activity. Kinoshita
et al. (2006) also showed that CWFI was associated with
higher vagal-related HRV indices compared with no
immersion or warm water immersion while either
breath-holding or snorkel-breathing. Finally, using phar-
macological blockade, Finley et al. (1979) showed that
CWFI-induced bradycardia at rest and during exercise
were related to an increased parasympathetic activity (i.e.,
inferred from slower HR recovery after parasympathetic
blockade).
While the stimulating effect of CWFI on parasympa-
thetic activity has been well described at rest (Finley et al.
1979; Hayashi et al. 1997; Paulev et al. 1990) and during
exercise (Finley et al. 1979; Smith et al. 1997), its effect on
post-exercise recovery has received little attention (Finley
et al. 1979). Concomitant activation of both autonomic
arms (i.e., CWFI-induced increased vagal activity com-
bined with post-exercise persistent sympathetic overactiv-
ity) may lead to a sympathovagal interaction, which might
contribute to a higher vagal activation compared with
resting conditions (Levy 1971; Tulppo et al. 1998).
Therefore, the aim of the present study was to examine
the effect of CWFI on parasympathetic reactivation fol-
lowing submaximal exercise performed under a heightened
sympathetic background (i.e., following maximal exer-
tion). We hypothesized that CWFI would trigger para-
sympathetic activity, as identified by faster post-exercise
parasympathetic reactivation compared with the control
condition.
123
Eur J Appl Physiol (2010) 108:599–606
Methods
Participants
Based on the assumption that a 2.1 ± 1.1 ms2 difference
in post-exercise high-frequency (HF) power (i.e., power
density of HRV in the HF ranges) and that 8 ± 5 beats/min
difference in HRR60s (i.e., absolute difference between the
final HR at exercise end and the HR recorded 60 s later) is
meaningful (Buchheit et al. 2007b), we used sigmaStat 3.11
Software (Systat software Inc., San Jose, CA, USA) to
determine that a sample size of at least eight participants
would provide a statistical power of 0.8 at an alpha level of
0.05. To further increase the statistical power of the study, we
eventually recruited 13 healthy males (21.6 ± 1.3 years;
1.80 ± 0.06 m; 76.1 ± 13.0 Kg, training 2–5 h per week)
to volunteer for the study after providing written informed
consent. Participants were all familiar with exercise testing,
not taking prescribed medications and presented with normal
levels of blood pressure and electrocardiographic patterns.
The study conformed to ethical guidelines outlined in the
Declaration of Helsinki and was approved by the local
human research ethics committee.
Exercise testing
Maximal graded aerobic test
Participants exercised on three separate occasions. First,
they performed an incremental and maximal intermittent
aerobic test (30-15IFT, Buchheit et al. 2009a) to determine
a reference velocity (VIFT) for the ongoing exercise session.
Second, 7 and 14 days later, they performed an intermittent
exercise sequence, comprised of a Wingate test (i.e., all-out
30-s cycling) followed after 5 min of passive seated
recovery, by a continuous submaximal running exercise
(i.e., 45% of VIFT). This exercise sequence was randomly
followed by 5 min passive (seated) recovery with either
CWFI or without (CON). Ambulatory HR recordings were
carried out 5 min before the Wingate test and throughout
the exercise and recovery periods. All participants were
asked to consume their last light meal at least 3 h before
each test session, and to refrain from smoking and con-
suming drinks containing caffeine.
Wingate test
All participants performed a Wingate test on a Monark
cycle ergometer (Monark 884E). Prior to the exercise test,
seat height was adjusted to accommodate the subject‘s
stature. They performed a 2-min warm-up pedaling at a
cadence of 80 rpm at a constant power output set at 50 W.
The warm-up included two brief (2–3 s) sprinting bouts.
Eur J Appl Physiol (2010) 108:599–606
Immediately after the warming-up period, the subjects
performed the 30-s Wingate test. Standard resistance
applied corresponded to 7.5% of the participant’s body
weight (Inbar et al. 1996). Participants were verbally
encouraged to perform maximally throughout the 30-s test.
The Wingate test was aimed at markedly increasing sym-
pathetic and reducing vagal activities (Goulopoulou et al.
2006) prior to the start of the submaximal exercise bout, so
that the confounding effect of sympathetic overactivity
(i.e., interaction; Levy 1971) on the post-exercise para-
sympathetic reactivation response to both recovery condi-
tions could be investigated. As well, analyzing autonomic
responses to submaximal, but not maximal exercise, was
designed to observe the influence of the presumed sym-
pathetic overactivity on post-exercise HRV indices, whilst
eliminating the potential confounding mechanical effects
of high minute ventilation on the HF power density (Blain
et al. 2005).
Submaximal exercise
The 5-min submaximal exercise was run at 45% of VIFT
without a warm-up. This intensity was chosen to ensure a
rapid return of HR to baseline levels following exercise
(Buchheit et al. 2009c), which is required for appropriate
short-term HRV analysis (i.e., steady-state signal). Run-
ning pace was governed by a prerecorded beep that soun-
ded at appropriate intervals in order to allow participants to
adjust their running speed as they passed through specific
zones of the field. Immediately at the end of exercise,
subjects were asked to recover passively in either CWFI or
CON conditions.
Cold water face immersion and control recovery
During both recovery conditions participants sat passively
and breathed through a short (25 cm; 80 ml) snorkel for
5 min wearing a nose clip. In CWFI, subjects remained
seated and bent their head forward into a basin of cold
water. The face immersed so that the forehead, eyes, and at
least two-thirds of both cheeks were submerged. Water
temperature was kept at 10–12°C.
Data measurements and analyses
Heart rate measurements
Heart rate was continuously recorded (s810 heart rate
monitor, Polar Electro, Kemple, Finland) during the exer-
cise sequence and the subsequent recovery phase. Recor-
ded data were downloaded on a computer using compatible
Polar software (Polar Precision Performance SW 5.20,
Polar Electro, Kempele, Finland). All irregular heartbeats
601
were automatically identified and replaced with interpo-
lated adjacent R–R interval values with the Polar Software
(Nunan et al. 2009).
Post-exercise HRR assessment
HRR was calculated by fitting the 5-min post-exercise R–R
data into a first-order exponential decay curve (Buchheit
et al. 2007a). A HR time constant (HRRs) was then produced
by modeling the resultant first 5 min of HR data using an
iterative technique (Sigmaplot 10; SPSS Science, Chicago,
IL) by the following equation: HR = HR0 ? HRamp
e(-T/HRRs), where HR0 is resting (final) HR, HRamp is
maximal HR (HRmax) - HR0, and T is time (s). HRR was
also assessed by calculating the absolute difference between
the final HR at exercise end and the HR recorded 60 s later
(HRR60s) (Buchheit et al. 2007a).
Short-term resting HRV analysis
HRV indices were calculated from the last three stationary
minutes of the 5-min recovery. Time domain indices were
natural logarithm of standard deviation of the normal R–R
intervals (Ln SDNN), natural logarithm of the square root
of the mean of the sum of the squares of differences
between adjacent normal R–R intervals (Ln rMSSD). Fast
Fourier Transform was conducted to obtain the power
densities in low- (LF; 0.04–0.15 Hz) and high-frequency
bands (HF; [0.15–0.4 Hz). HRV indices were calculated
directly with the accompanying Polar software (Polar
Precision Performance SW 5.20, Polar Electro, Kempele,
Finland), which has been shown to provide accurate mea-
surements for short term HRV analysis (Nunan et al. 2009).
Although respiratory rate is often controlled in HRV
studies, we chose not to control respiratory rate in our
participants because we did not want to perturb the natural
return of HR to baseline levels. Nevertheless, respiratory
rate was always in the HF range ([0.15–0.50 Hz) and did
not differ significantly in both recovery conditions. The Ln
SDNN, Ln rMSSD, natural logarithm of LF (LnLF), and
HF (LnHF) power density, the normalized HF power
(i.e., HFnu = HF/(LF ? HF)) were retained for statistical
analysis.
Time-varying vagal-related HRV index
A time-varying vagal-related index, the square root of the
mean of the sum of the squares of differences between
adjacent normal R–R intervals (rMSSD), was calculated
for each of the 30-s segments of recovery (rMSSD30s) for
the two 5-min recovery conditions. (Goldberger et al.
2006). Data were median-filtered in order to smooth out
transient outliers in the HRV plots (HRV vs. time in
123
602
recovery). The first and last values were not median filtered
(Goldberger et al. 2006). rMSSD30s was computed for the
5-min of recovery in the CWFI and CON conditions.
Statistical analyses
The distribution of each variable was examined with the
Shapiro–Wilk normality test. Homogeneity of variance
was verified by a Levene test. When data were skewed
(e.g., HF power density), data were transformed by taking
the natural logarithm to allow parametric statistical
comparisons that assume a normal distribution. Difference
between recovery methods in HRV and HRR indices were
compared using a paired t test. Time-varying rMSSD30s
was analyzed using a two-way ANOVA for repeated
measures, with one within (‘time’, i.e., ten consecutive
measurements during recovery) and one between factor
(‘condition’, i.e., CWFI vs. CON). When a significant
interaction was noted, a Bonferroni’s post hoc tests were
conducted to delineate the main effects and/or interactions
of the recovery condition and time during recovery.
Traditional statistical analyses (paired t test and two-way
ANOVA) were performed using SigmaStat software
(SigmaStat 3.11, Systat software inc., San Jose, CA,
USA). Significance level was set at P \ 0.05. If no sig-
nificant effects were observed, but a tendency towards
significance (P \ 0.1) was apparent, then adjusted effect
sizes (aES) were calculated (Cohen 1988). If there was at
least a moderate aES ([0.50), but the statistical power
was low, the likelihood of a type II error was noted. aES
formula is as follows:
ES
aES ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð1  rÞ
where r stands for correlation coefficient between before
and after data and ES stands for effect size:
ma  mb
ES ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2
ffi
ðSDa Þðna 1ÞþðSDb Þðnb 1Þ
ðna þnb 2Þ
where b stands for CWFI, a for CON, m for group mean,
SD for standard deviation, n for group subject size. The
scale proposed by Cohen (1988) was used for interpreta-
tion. The magnitude of the difference was considered either
small (aES [ 0.2), moderate ([0.5), or large ([0.8).
Because traditional statistical analyses have been shown as
being not well adapted in sports medicine and exercise
science research (Hopkins et al. 2009), we used the Hop-
kins spreadsheet (Hopkins 2007) to assess the likelihood
that the true effect was beneficial (substantially positive),
trivial, or harmful (substantially negative). The smallest
worthwhile change was calculated as the pooled SD 9 0.2
(based on Cohen’s effect size principle). The qualitative
123
Eur J Appl Physiol (2010) 108:599–606
terms and the default values are as follows: most unlikely,
\0.5%; very unlikely, 0.5–5%; unlikely, 5–25%; possibly,
25–75%; likely, 75–95%; very likely, 95–99.5%; and most
likely, [99.5%. If the chance of having a beneficial or
harmful effect were both [5%, the true difference was
assessed as unclear (Batterham and Hopkins 2006; Hopkins
et al. 2009). Data in text are presented as means and
standard deviations (±SD), except in Fig. 3, where data are
presented as means ± standard errors (SE) for clarity.
Results
Maximal graded aerobic test and Wingate test
Mean VIFT was 19 ± 1 Km h-1. Mean maximal power
output for the Wingate test was similar for CON and CWFI
conditions (730 ± 113 vs. 733 ± 121 W for CON and
CWFI, respectively, P = 0.84).
Parasympathetic reactivation indices
Post-exercise heart rate recovery indices
HRR indices are reported in Table 1 and their qualitative
analyses are presented in Fig. 1. We found a faster HRR60s
for CWFI compared with the CON condition (P = 0.002),
and a tendency towards a shorter HRRs was noted in the
CWFI condition (P = 0.07, aES = 0.76). Qualitative
analysis supported the beneficial effect of CWFI on post-
exercise parasympathetic reactivation with a ‘‘very likely’’
(99% beneficial) for HRR60s and ‘‘likely’’ (86% beneficial)
effect for HRRs.
Post-exercise heart rate variability indices
Visual examination of individual HR traces for both
recovery conditions confirmed the stableness of the 3-min
analyzed period for spectral HRV analyses. HRV indices
are reported in Table 1 and their qualitative analyses are
represented in Fig. 1. Vagal related indices were increased
in CWFI compared to CON (LnHF, P \ 0.001; Ln rMSSD
P = 0.026, HFnu P = 0.002). The sympathovagal balance
was shifted toward a greater parasympathetic dominance in
CWFI compared to CON (LF/HF, P = 0.002). Qualitative
analysis of the beneficial effect of CWFI on post-exercise
parasympathetic reactivation supports these statistics by
revealing ‘‘very likely’’ (95% beneficial) effects for LnHF,
‘‘likely’’ (82% beneficial) effects for Ln rMSSD, ‘‘most
likely’’ (100% beneficial) effects for HFnu, and ‘‘most
likely’’ (100% beneficial) effects for LF/HF. No differ-
ences were noted between Ln SDNN (unclear) and LnLF
(unclear) in CWFI compared with CON conditions.
Eur J Appl Physiol (2010) 108:599–606 603

Table 1 Heart rate variability (HRV) and heart rate recovery (HRR)
indices following the submaximal exercise in control and cold water HRR60s most likley beneficial
face immersion conditions likely beneficial
HRRτ

CON CWFI Ln SDNN unclear

HRR indices Ln rMSSD likely beneficial

HRR60s 30 ± 8 43 ± 13* LnHF most likley beneficial

HRRs 48 ± 17 37 ± 17  LnLF unclear

HRV indices
HFnu most likley beneficial
Ln SDNN (ms) 3.08 ± 0.45 3.26 ± 0.65
LF/HF most likley beneficial
Ln rMSSD (ms) 1.87 ± 0.60 2.36 ± 0.89*
2
LnHF (ms ) 2.84 ± 1.40 4.34 ± 1.69**
-100 0 100 200
LnLF (ms2) 4.77 ± 1.22 5.18 ± 1.61 Difference (%)
HFnu 0.14 ± 0.06 0.32 ± 0.16*
LF/HF 7.56 ± 3.39 3.00 ± 1.98* Fig. 1 Quantitative statistical analysis showing differences between
cold water face immersion and control recovery conditions on HR-
Mean ± SD of absolute difference between the final heart rate at related indices. Absolute difference between the final heart rate at
exercise end and the heart rate recorded 60 s later (HRR60s), time exercise end and the heart rate recorded 60 s later (HRR60s), the time
constant of heart rate decay (HRRs), the natural logarithm of the constant of heart rate decay (HRRs), the natural logarithm of the
standard deviation of R–R intervals (Ln SDNN), the natural logarithm standard deviation of R–R intervals (Ln SDNN), the natural logarithm
of the root square of the mean squared differences of successive R–R of the root square of the mean squared differences of successive R–R
intervals (Ln rMSSD), the natural logarithm of the high-frequency intervals (Ln rMSSD), the natural logarithm of the high-frequency
power (LnHF), the natural logarithm of the low-frequency power power (LnHF), the natural logarithm of low-frequency power (LnLF),
(LnLF), normalized unit in high-frequency range (HFnu), and sym- normalized unit in high-frequency range (HFnu),and sympathovagal
pathovagal balance (LF/HF). HRV were analyzed during the last balance (LF/HF). Bars indicate uncertainty in the true mean changes
3 min of the 5-min recovery in control (CON) and cold water face with 90% confidence intervals. Trivial area was calculated as the
immersion (CWFI) conditions smallest worthwhile change (see ‘‘Methods’’)
* P \ 0.05 versus control
** P \ 0.001 versus control 600
 
Differences versus control with adjusted effect size considered as
moderate
550
R-R intervals (ms)

Poincaré plots were drawn for all participants and were

comet shaped in both recovery conditions. R–R intervals 500
and associated Poincaré plots in a representative participant
during the 3-min recovery periods under both CWFI and
450
CON conditions are illustrated in Fig. 2.
CWFI
CON
Post-exercise time varying vagal-related HRV index 400
0 20 40 60 80 100 120 140 160 180 200
Time (s)
Figure 3 illustrates the time course of the time-varying
CWFI CON
rMSSD30s in both recovery conditions. We found a recovery 540 540
R-Rn+1 (ms)
R-Rn+1 (ms)

condition effect (P = 0.034) and time effect (P \ 0.001), 520 520

without a significant ‘time 9 condition’ interaction
500 500
(P = 0.12). Although this interaction was not significant,
Bonferroni’s post hoc test showed within-condition differ- 480 480

ences for ‘time’ (i.e., rMSSD30s at 30 s vs. 60 s). 480 500 520 540 480 500 520 540
R-Rn (ms) R-Rn (ms)

Fig. 2 R–R intervals and associated Poincaré plots in a representa-

Discussion
The present study examined for the first time the effect of
cold water face immersion (10–12°C) on post-exercise
parasympathetic reactivation, as inferred from heart rate
variability and heart rate recovery indices. In accordance
with our hypotheses, results showed that cold water face
tive subject during the 3-min periods under cold water face immersion
(CWFI) and Control (CON) recovery conditions. Each R–R interval
(R–Rn?1) is plotted as a function of previous R–R interval (R–Rn)
immersion increased vagal-related indices of HRV, and
was therefore shown as a potent trigger of parasympathetic
activity immediately after exercise.

123
604
30
25
* profound effect on thermolysis (Pretorius et al. 2006).
rMSSD30s (ms)
20
15
10
5 post-exercise parasympathetic reactivation
0
0 50 100 150 200 250 300 350
Time (s)
Fig. 3 Average (±SE) of the square root of the mean of the sum of
the squares of differences between adjacent normal R–R intervals
measured on successive 30-s segments (rMSSD30s) during the 5-min
post-exercise recovery period, as calculated for participants with cold
water face immersion (filled triangles) or control (open triangles).
Circles on symbols denote significant within-condition difference for
‘time’ (vs. rMSSD30s at 30 s after the end of exercise). *Significant
main effect for recovery condition, i.e., difference versus control
condition (P \ 0.05)
Effect of cold water face immersion on immediate
post-exercise heart rate recovery
Wingate test performance was as expected (Popadic
Gacesa et al. 2009) and was comparable for both recovery
conditions. This implies that the presumed oversympa-
thetic activity at the start of the submaximal runs was also
likely to be similar for both conditions. Our results showed
a most likely increase in HRR60s and a likely shortening in
HRRs following cold water face immersion (Table 1 and
Fig. 1), which was likely related to the activation of cold
trigemino-cardiac reflex receptors in the face, known to
trigger parasympathetic activity. This finding is in accor-
dance with the bradycardia effect of cold water face
immersion noted in several studies at rest (Finley et al.
1979; Hayashi et al. 1997; Kinoshita et al. 2006) and
during exercise (Finley et al. 1979; Smith et al. 1997). It
is, however, worth noting that this presumed effect on
cardiac autonomic control is thought to be mediated by
trigeminal nerve activation, without any evident implica-
tion of the baroreflex loop (Khurana and Wu 2006). This
confirms the assumption that HRR is directly related to
vagal tone, and perhaps less to its modulation (Buchheit
et al. 2007a; Dewland et al. 2007), which was less likely
to have been modified in the present cold face water
immersion condition (e.g., no apparent change in central
blood volume). Another possible mechanism that could
explain the faster HRR in the cold water face immersion
condition may be related to the direct (non-autonomic)
123
Eur J Appl Physiol (2010) 108:599–606
effect that core temperature has on HR (Buchheit and
Laursen 2009; Gorman and Proppe 1982). Indeed, whole
head cold water immersion has been shown to have a
While core temperature was not measured in the present
study, it is possible that cold water face immersion
might have accelerated heat dissipation and consequently
enhanced HRR.
Effect of cold water face immersion on immediate
Limitations to HRV indices are well known. While HRV
offers a qualitative marker of cardiac parasympathetic
regulation, care should nevertheless be taken when inter-
preting sympathovagal balance indices (i.e., LF/HF),
because the LF component reflects some unknown inter-
actions between sympathetic and parasympathetic inputs
(Task Force 1996). Vagal related indices calculated under
steady-state condition (i.e., LnHF, Ln rMSSD) were most
likely and very likely (respectively) increased and most
likely shifted towards an increased parasympathetic pre-
dominance (Table 1 and Fig. 1). Analysis of the time-
varying vagal-related index (i.e., rMSSD30s) revealed also
a faster (i.e., significant difference noted at 60 and 90 s in
cold water face immersion condition compared with 30 s,
despite no difference in control condition; Fig. 3) and
greater parasympathetic reactivation in the cold water face
immersion condition (i.e., smaller P values). A co-activa-
tion of cardiovagal and sympathetic efferents during cold
water face immersion has been previously reported
(Khurana 2007), and is likely a result of simultaneous cold
water face immersion-induced increases in vagal activity
due to trigeminal cutaneous receptors stimulation (Eckberg
et al. 1984; Finley et al. 1979; Khurana and Wu 2006;
Kinoshita et al. 2006; LeBlanc et al. 1976; Paulev et al. 1990)
and post-exercise elevated sympathetic levels consecutive
to the previous supramaximal exercise (Goulopoulou et al.
2006). An increased vagal activation can be observed as a
result of a sympathovagal interaction, which occurs when
both arms of the autonomic system are highly activated
(Levy 1971; Tulppo et al. 1998). Conversely, sympathetic
activation can also attenuate the vagal influence on HR
(Miyamoto et al. 2003). Since normal comet-shaped scatter
plots (and no torpedo-shaped or parabola-like plots) were
observed for all subjects in both conditions (Fig. 2, bottom)
(Tulppo et al. 1998), the occurrence of an eventual sym-
pathovagal interaction was unlikely. Although the inter-
pretation of sympathetic activity level from HRV indices
should be viewed with caution (Task Force 1996), the
similar LnLF power values for both recovery conditions
suggests that cold water face immersion following exercise
may not additionally activate the sympathetic system. It is
Eur J Appl Physiol (2010) 108:599–606
also possible that water temperature was not cold enough to
activate cold receptors responsible for sympathetic over-
drive (Tipton 1989). We can therefore postulate that the
increase in vagal-related HRV indices for the cold water
face immersion condition, as implemented in the present
study, is likely related to exclusive changes in parasym-
pathetic activity. Since the parasympathetic effect of cold
water face immersion is thought to be non-baroreflex
mediated (Khurana and Wu 2006), the changes observed in
vagal-related HRV indices, generally thought to essentially
rely on parasympathetic modulation (Buchheit et al. 2007a;
Hedman et al. 1995), are surprising. Baroreflex function
was, however, not investigated in the present experimen-
tation, and it is possible that the respective link between
either parasympathetic tone or modulation and HRR or
HRV are not as evident as previously thought (Buchheit
et al. 2007a; Hedman et al. 1995). Possible changes in
respiratory pattern (e.g., changes in tidal volume or intra
thoracic pressure), not measured here, may have been
responsible for the changes (Eckberg 2003) observed in the
cold water face immersion condition.
Implications for the clinical use of cold water face
immersion during exercise recovery
Since it is well established that lowered post-exercise
vagal activity is associated with sudden cardiac death
(Billman 2002), our findings encourage the use of cold
water face immersion to decrease cardiac work and lower
the risk of a cardiovascular event, at least in the acute
setting. Although evaluation of the influence that cold
water face immersion has on autonomic function is
important in sedentary individuals and patients, we pur-
posefully recruited moderately trained subjects because
we expected that they would be able to handle the
demands of the present experimentation. Trained subjects
often display high parasympathetic responses during
exercise recovery (Buchheit and Gindre 2006), so the
present results should be viewed with caution when
inferences are being made into the effect that these
interventions might have in a patient population. Whether
individuals with low physical activity and/or parasympa-
thetic levels show a similar response to cold water face
immersion also needs confirming. At last, since present
measurements were restricted to the immersion period,
whether the observed beneficial effect of cold water face
immersion on cardiac parasympathetic function is likely
to be prolonged afterward is not known. Further research
evaluating the effect of cold water face immersion on
HR-derived indices during the post-immersion period
might help in defining the range of (clinical) applications
for this method.
605
Conclusion
To conclude, while investigating for the first time the effect
of cold water face immersion on vagal-related HR-derived
indices, we showed that the application of a cold water
immersion to the face immediately after exercise appears to
be a simple, inexpensive, non-invasive, and effective
means of immediately accelerating post-exercise para-
sympathetic reactivation.
Acknowledgments The authors thank Bachar Haidar for his helpful
assistance with data collection and the participants for their enthusi-
astic collaboration.
References
Batterham AM, Hopkins WG (2006) Making meaningful inferences
about magnitudes. Int J Sports Physiol Perform 1:50–57
Billman GE (2002) Aerobic exercise conditioning: a nonpharmaco-
logical antiarrhythmic intervention. J Appl Physiol 92:446–454
Billman GE (2006) Heart rate response to onset of exercise: evidence
for enhanced cardiac sympathetic activity in animals susceptible
to ventricular fibrillation. Am J Physiol Heart Circ Physiol
291:H429–H435
Blain G, Meste O, Bermon S (2005) Influences of breathing patterns
on respiratory sinus arrhythmia in humans during exercise. Am J
Physiol Heart Circ Physiol 288:H887–H895
Buchheit M, Gindre C (2006) Cardiac parasympathetic regulation:
respective associations with cardiorespiratory fitness and training
load. Am J Physiol Heart Circ Physiol 291:H451–H458
Buchheit M, Laursen PB (2009) Treatment of hyperthermia: is
assessment of cooling efficiency enough? Exp Physiol 94:627–
629
Buchheit M, Papelier Y, Laursen PB, Ahmaidi S (2007a) Noninvasive
assessment of cardiac parasympathetic function: post-exercise
heart rate recovery or heart rate variability? Am J Physiol Heart
Circ Physiol 23:H8–H10
Buchheit M, Laursen PB, Ahmaidi S (2007b) Parasympathetic
reactivation after repeated sprint exercise. Am J Physiol Heart
Circ Physiol 293:H133–H141
Buchheit M, Al Haddad H, Millet GP, Lepretre PM, Newton M,
Ahmaidi S (2009a) Cardiorespiratory and cardiac autonomic
responses to 30–15 intermittent fitness test in team sport players.
J Strength Cond Res 23:93–100
Buchheit M, Peiffer JJ, Abbiss CR, Laursen PB (2009b) Effect of cold
water immersion on postexercise parasympathetic reactivation.
Am J Physiol Heart Circ Physiol 296:H421–H427
Buchheit M, Al Haddad H, Laursen PB, Ahmaidi S (2009c) Effect of
body posture on postexercise parasympathetic reactivation in
men. Exp Physiol 94:795–804
Cohen J (1988) Statistical power analysis for the behavioral sciences.
Lawrence Erlbaum, Hillsdale
Cole CR, Blackstone EH, Pashkow FJ, Snader CE, Lauer MS (1999)
Heart-rate recovery immediately after exercise as a predictor of
mortality. N Engl J Med 341:1351–1357
Dewland TA, Androne AS, Lee FA, Lampert RJ, Katz SD (2007)
Effect of acetylcholinesterase inhibition with pyridostigmine on
cardiac parasympathetic function in sedentary adults and trained
athletes. Am J Physiol Heart Circ Physiol 293:H86–H92
Eckberg DL (2003) The human respiratory gate. J Physiol 548:339–
352
123
606
Eckberg DL, Mohanty SK, Raczkowska M (1984) Trigeminal-
baroreceptor reflex interactions modulate human cardiac vagal
efferent activity. J Physiol 347:75–83
Finley JP, Bonet JF, Waxman MB (1979) Autonomic pathways
responsible for bradycardia on facial immersion. J Appl Physiol
47:1218–1222
Goldberger JJ, Challapalli S, Tung R, Parker MA, Kadish AH (2001)
Relationship of heart rate variability to parasympathetic effect.
Circulation 103:1977–1983
Goldberger JJ, Le FK, Lahiri M, Kannankeril PJ, Ng J, Kadish AH
(2006) Assessment of parasympathetic reactivation after exer-
cise. Am J Physiol Heart Circ Physiol 290:H2446–H2452
Gorman AJ, Proppe DW (1982) Influence of heat stress on arterial
baroreflex control of heart rate in the baboon. Circ Res 51:73–82
Goulopoulou S, Heffernan KS, Fernhall B, Yates G, Baxter-Jones
AD, Unnithan VB (2006) Heart rate variability during recovery
from a Wingate test in adolescent males. Med Sci Sports Exerc
38:875–881
Hayashi N, Ishihara M, Tanaka A, Osumi T, Yoshida T (1997) Face
immersion increases vagal activity as assessed by heart rate
variability. Eur J Appl Physiol Occup Physiol 76:394–399
Hedman AE, Hartikainen JE, Tahvanainen KU, Hakumaki MO
(1995) The high frequency component of heart rate variability
reflects cardiac parasympathetic modulation rather than para-
sympathetic ‘tone’. Acta Physiol Scand 155:267–273
Hopkins WG (2007) A spreadsheet for deriving a confidence interval,
mechanistic inference and clinical inference from a p value.
Sportscience 11:16–20
Hopkins WG, Marshall SW, Batterham AM, Hanin J (2009)
Progressive statistics for studies in sports medicine and exercise
science. Med Sci Sports Exerc 41:3–13
Imai K, Sato H, Hori M, Kusuoka H, Ozaki H, Yokoyama H, Takeda
H, Inoue M, Kamada T (1994) Vagally mediated heart rate
recovery after exercise is accelerated in athletes but blunted in
patients with chronic heart failure. J Am Coll Cardiol 24:1529–
1535
Inbar O, Bar-Or O, Skinner JS (1996) The Wingate anaerobic test.
Human Kinetics, Champaign, IL
Khurana RK (2007) Cold face test: adrenergic phase. Clin Auton Res
17:211–216
Khurana RK, Wu R (2006) The cold face test: a non-baroreflex
mediated test of cardiac vagal function. Clin Auton Res 16:202–
207
Kinoshita T, Nagata S, Baba R, Kohmoto T, Iwagaki S (2006) Cold-
water face immersion per se elicits cardiac parasympathetic
activity. Circ J 70:773–776
LeBlanc J, Blais B, Barabe B, Cote J (1976) Effects of temperature
and wind on facial temperature, heart rate, and sensation. J Appl
Physiol 40:127–131
123
Eur J Appl Physiol (2010) 108:599–606
Levy MN (1971) Sympathetic–parasympathetic interactions in the
heart. Circ Res 29:437–445
Miyamoto T, Kawada T, Takaki H, Inagaki M, Yanagiya Y, Jin Y,
Sugimachi M, Sunagawa K (2003) High plasma norepinephrine
attenuates the dynamic heart rate response to vagal stimulation.
Am J Physiol Heart Circ Physiol 284:H2412–H2418
Miyamoto T, Oshima Y, Ikuta K, Kinoshita H (2006) The heart rate
increase at the onset of high-work intensity exercise is acceler-
ated by central blood volume loading. Eur J Appl Physiol 96:86–
96
Mourot L, Bouhaddi M, Gandelin E, Cappelle S, Dumoulin G, Wolf
JP, Rouillon JD, Regnard J (2008) Cardiovascular autonomic
control during short-term thermoneutral and cool head-out
immersion. Aviat Space Environ Med 79:14–20
Nunan D, Donovan G, Jakovljevic DG, Hodges LD, Sandercock GR,
Brodie DA (2009) Validity and reliability of short-term heart-
rate variability from the Polar S810. Med Sci Sports Exerc
41:243–250
Paulev PE, Pokorski M, Honda Y, Ahn B, Masuda A, Kobayashi T,
Nishibayashi Y, Sakakibara Y, Tanaka M, Nakamura W (1990)
Facial cold receptors and the survival reflex ‘‘diving bradycar-
dia’’ in man. Jpn J Physiol 40:701–712
Popadic Gacesa JZ, Barak OF, Grujic NG (2009) Maximal anaerobic
power test in athletes of different sport disciplines. J Strength
Cond Res 23:751–755
Pretorius T, Bristow GK, Steinman AM, Giesbrecht GG (2006)
Thermal effects of whole head submersion in cold water on
nonshivering humans. J Appl Physiol 101:669–675
Savin WM, Davidson DM, Haskell WL (1982) Autonomic contribu-
tion to heart rate recovery from exercise in humans. J Appl
Physiol 53:1572–1575
Smith JC, Stephens DP, Winchester PK, Williamson JW (1997)
Facial cooling-induced bradycardia: attenuating effect of central
command at exercise onset. Med Sci Sports Exerc 29:320–325
Smith LL, Kukielka M, Billman GE (2005) Heart rate recovery after
exercise: a predictor of ventricular fibrillation susceptibility after
myocardial infarction. Am J Physiol Heart Circ Physiol
288:H1763–H1769
Task Force of the European Society of Cardiology and the North
American Society of Pacing and Electrophysiology (1996) Heart
rate variability, standards of measurement, physiological inter-
pretation, and clinical use. Circulation 93:1043–1065
Tipton MJ (1989) The initial responses to cold-water immersion in
man. Clin Sci (Lond) 77:581–588
Tulppo MP, Makikallio TH, Seppanen T, Airaksinen JK, Huikuri HV
(1998) Heart rate dynamics during accentuated sympathovagal
interaction. Am J Physiol 274:H810–H816