Keywords
apnea performance - freediving - cognition
Introduction
Competitive breath-hold (BH) diving is a recent sport in which the aim is to achieve
the longest possible BH without any movement (Static Breath-Hold discipline,
SBH) or the longest distance in a swimming pool (Dynamic Breath-Hold
discipline, DYNBH) with or without fins or as deep as possible in the sea
(constant weight) also with or without fins. The regular new world records in all
the BH disciplines bear witness to the improved training techniques and increased
training load of these top athletes [1].
Breath-hold training enables athletes to be very economical with oxygen thanks to
an
effective diving response, characterized mainly by bradycardia with peripheral
vasoconstriction [1]
[2]. Precise control and manipulation of
the training load are necessary to adjust the stress applied to the athlete at the
individual level [3]. This training load
can be described as being internal (all the psychophysiological responses) and
external (the physical work prescribe in the training plan) [4]. External and internal loads are
therefore linked by a causal relationship [5]. The Training Impulse (TRIMP) is one of the indices for integrating
external and internal training loads [6]. It combines the heart rate (HR) reserve method (the internal load), the
duration of training (the external load) and a weighting factor [6]. The Edward TRIMP model (eTRIMP) is an
alternative of TRIMP for quantifying the physiological impact of training by using
the accumulated duration in five arbitrary HR zones multiplied by a weighting factor
[7]. Such methods based on eTRIMP
cannot be used for Breath-Hold Divers (BHDs) because HR decreases during BH in a
non-linear way with the exercise intensity [8]. Nevertheless, it is important to define the BH training load because
such efforts results in significant desaturation (below 50%), which can lead to loss
of consciousness (LOC) [9] or potential
brain lesions [10]
[11]. The impact on cognitive function is
less certain. Billaut et al. [12]
suggest an impairment of executive functions or attention deficit characterized by
poorer Stroop test scores in elite BHDs compared to novice BHDs or control subjects,
while Ridgway et al. [13] do not observe
any deleterious effects. Apart from the fact that the vast majority of BHDs are not
high-level, most of them train in the pool and some in the sea, with repetitive BHs
during training that generally last between one and three hours. To date, the
relationship between the impact of BH training load and its different training
zones, on cognitive performance, especially memory, remains an open question that
needs to be investigated in an ecological (real-life) situation.
Our first aim was to develop several specific indices to measure the BH training
load. Secondly, we investigated the relationship between these load levels during
ecological BH pool training and memory in well-trained BHDs. We hypothesized that
i)
our new index would provide a reliable training load estimation offering coaches a
valuable tool for individualizing and optimizing BHD training programs and ii) the
training load would be negatively correlated with memory test performances.
Materials and Methods
Participants
Eighteen participants took part in this study (14 men and 4 women) with a mean
age of 35.8±6.6 years and BH diving practice for 5.3±4.5 years. Their personal
bests in SBH and DYNBH with bifins were 294.2±65.1 seconds
and 87.5±31.1 meters, respectively. Of the 18 BHDs tested, six experienced at
least one LOC, with an average of two LOC (ranging from 1 to 6). All
participants were non-smokers, not dependent on alcohol or drugs, and had no
contraindication to BH training. The local ethics committee approved the entire
protocol, and informed written consent was obtained from all participants.
Design
After answering standard questions about their BH diving
history (personal bests, years of BH practice, number of LOC); they took three
different memory tests 20 minutes before and 10 minutes after their usual BH
training in a pool. The BHDs are equipped with a HR monitor (Polar© H10) during
their BH training in the pool. The HRs are then collected after the training
using the Polar Beat Software© and exported as an Excel file for further
analysis. An experimenter recorded the beginnings and endings of BHs throughout
the training while remaining out of the water. At the end of the training
course, the BHDs rated their perceived exertion on a Borg scale (RPE) and their
sensation of breathlessness on a dyspnea scale (RPD) from 0 to 10. The duration
and configuration of BH training were the same for all tested BHDs and start for
all at 19:00 hours ([Fig. 1]). The
BH training took place as follows: (i) first, a warm-up period involving free
swimming; (ii) followed by hypercapnic BH exercises (short, moderately intense
efforts with short recovery) combined with technical exercises to improve
hydrodynamics; (iii) and finally, hypoxic BH exercises (long efforts with almost
complete recovery) to train for the lack of oxygen. All BHDs’ dive durations and
distances were summed and averaged over the entire training. These parameters
were used to calculate the average BH speed (m/s). Bradycardia was calculated as
the difference in HR between the start of BH (HRstart) and the lowest
HR value recorded during BH (HRmin) and expressed as a percentage
(%HR). A reference value (Rv) was used to compare our training load indices and
calculated as the average heart rate (HRavg) normalized by
HRmin during all effective BHs. This Rv is used as an indicator
of intensity, a lower value indicating more intense bradycardia and possibly a
more advanced state of hypoxia.
Fig. 1 Typical exercise that makes up the BHD training and effect
on heart rate. Hypercapnic exercises are short repetitions with very
little recovery increasing carbon dioxide levels in the body. Hypoxic
exercises are longer repetitions with more recovery, which leads to a
decrease in oxygen levels in the body. HR: heart rate; BPM: beat per
minute.
We also calculated different load indices, considering both the
internal and/or external load during their BH training. We used the relationship
between HR, speed and RPE with exertion intensity as representative of the level
of hypoxia during BH [14]
[15]. The first equation (eq. 1)
represents the product of HRavg expressed as percentage of the
HRstart (%HRavg), by the average BH duration
(BHdavg) and the total number of BHs (n).
The second equation (eq. 2),
represents the product between the average speed (Vavg) by the total
BH duration during training (BHdtot) and the total number of BHs (n).
The third equation
(eq. 3), represents the product between the RPE and the total BH duration during
training (BHdtot). This equation quantifies the training load in a
subjective way [16].
The fourth equation (eq. 4)
represents the sum of the ratios of the starting HR (HRstart) and the
minimum HR (HRmin) for each BH, and standardizes HRmin
values according to the theoretical maximum HR (HRmaxth) [17] to obtain %HRmin (eq.
4a). Then, this ratio is multiplied by the BH duration (BHd) (eq. 4b).
We propose to call
the eq. 4b, the “apnea” TRIMP (aTRIMP). The aTRIMP will enable us to quantify
the impact of BH training, taking into account both bradycardia and intensity of
effort during DYNBH.
This equation (eq. 4b) is used for each BH
to establish intensity zones from 1 to 6 and is calculated as
follows:
Finally, the percentage
of time spent in each zone is determined by dividing the total BH duration in a
specific zone by the BHdtot (eq. 4c).
The number of significant
figures before the decimal point in the equations used to calculate the training
load were adjusted to ensure consistent values and make comparisons easier.
Before and within ten minutes after their BH training, the BHDs performed the
same memory tests. Three tests classically used to measure the different
components of memory were chosen: the phonemic fluency test [18], the categorical verbal fluency
test and the Mnemonic Similarity Task (MST) test [19]. The phonemic verbal fluency test
evaluates semantic memory and executive function. In this test, participants
have one minute to say as many words as possible beginning with the letter “P”,
avoiding proper nouns. The number of legal words generated in one minute was
recorded and used as a test score. The categorical verbal fluency test or
semantic fluency is used to evaluate the integrity of semantic memory.
Participants had one minute to list as many animal names as possible. The score
of this test corresponded to the number of words listed by the participant.
Finally, the Mnemonic Similarity Task measures a behavioral proxy of pattern
separation, which consists in the assignment of non-overlapping neural
activation patterns to similar memory representations by the hippocampus. In
other words, pattern separation allows to make sure that similar events remain
distinguishable in memory. We used a version of the MST that consists of an
initial incidental encoding phase in which participants have to perform a
semantic judgment task on pictures, followed by a test phase evaluating
different aspects of memory [19]
[20]. Pattern
separation (PS) index was defined as the percentage of correct responses
identifying images similar to images shown in the encoding phase (“lure”
images), minus the percentage of wrong responses when the participants
incorrectly identified lure images as images identical to the ones of the first
phase (“target” images). Moreover, the item memory index (IM), which measures
recognition memory, was defined as the percentage of correctly identified
“target” images minus the percentage of wrong responses when the participants
incorrectly identified “target” images as new images (“foil” images) [19]
[20].
Statistical Analyses
In order to be able to calculate a priori the number of subjects required for
this study, we chose as the main criterion the variation in HR during dynamic BH
in BHDs (delta HR %=0.33±0.4) according to the study by Andersson et al. [21]. For a type I error rate of 5%,
the inclusion of 11 participants provides 80% power to detect the effect of BH
on heart rate (GPower v3.1.9.2). Given the possibility that some participants
may not complete the intervention period, it was planned to include 18 BHDs to
ensure the above power level was achieved, guaranteeing the minimum sample size
of 11 participants. Dependent (HR, RPE, swimming speed, training load equation
results, memory test results) and independent (Total BH distance, training
duration) variables were tested for normality of distribution with the
Shapiro-Wilk’s test. The z-score of all memory tests were calculated using norms
to assess whether the participants had any deficits in these cognitive functions
beforehand [18]. The z-score will
also be compared to 40 healthy control participants (20 women and 21 men, mean
age 30.42±12.34 years, range=20.00–56.00) without a history of BH training,
neurological or psychiatric issues. A Pearson correlation analysis was performed
to investigate the association between our reference value and the training load
indices, as well as training parameters and the training load indices. The
optimal training load index will be determined as the one exhibiting the highest
correlation with our reference value. These correlations were additionally
employed to investigate potential relationships between memory tests and both
training parameters and training load indices. Statistical analyses were
performed using JASP Team (2023 Version 0.18.1). The level of statistical
significance was defined at 95% and the probability values of p<0.05.
Results
BH training parameters and training load indices results are presented as means±SD
and coefficient of variation (CV) in [Table
1]. BHDs spent an average of 80% (±11%) of the total BH duration during
training in zone 3 (moderate), 19% (±10%) in zone 4 (vigorous) and 1% (±1%) in zone
5 (very hard) of the aTRIMP. Two significant positive correlations were observed
between our four training load equations and the Rv: equation 1 (r=0.507, p<0.05)
([Fig. 2a]) and equation 4 (aTRIMP)
(r=0.652, p<0.01) ([Fig. 2b]).
Correlation analyses between the training load equations showed only a significant
correlation between equation 1 and aTRIMP (r=0.671, p<0.01). The total BH
duration showed a significant negative correlation with the HRmin during
training (r=− 0.610, p<0.01). Of the four equations, only equation 3 showed a
positive correlation with RPE (r=0.840, p<0.001) and RPD (r=0.806,
p<0.001).
Fig. 2
(a) Correlation between equation 1 and the reference value (r=0.507,
p<0.05). (b) Correlation between aTRIMP and the reference value
(r=0.652, p<0.01).
Table 1 Training parameters of the breath-holding (BH)
training.
|
BHDS (n=18)
|
Mean±SD
|
CV
|
|
BHdtot
|
(min)
|
20.2±3.4
|
0.171
|
|
(%)
|
22.6±3.8
|
0.169
|
|
Total number of BH (n)
|
33.1±7.3
|
0.219
|
|
Bradycardia
|
(beats/min)
|
9.7±5.7
|
0.581
|
|
(%)
|
10.4±5.5
|
0.534
|
|
BHdavg (sec.)
|
37.9±7.2
|
0.190
|
|
Total BH distance (m)
|
761.7±217.8
|
0.286
|
|
Vavg (m.s-1)
|
0.8±0.1
|
0.192
|
|
RPE
|
3.7±0.9
|
0.257
|
|
RPD
|
3.7±1.0
|
0.266
|
|
Equation 1
|
127±80*
|
0.629
|
|
Equation 2
|
146±65
|
0.448
|
|
Equation 3
|
448±15
|
0.332
|
|
Equation 4 (aTRIMP)
|
253±13**
|
0.497
|
|
Reference value
|
158±13
|
0.080
|
BHDs: breath-hold divers; RPE: rate of perceived exertion; RPD: rate of
perceived dyspnea; Vavg: average BH speed; HRavg:
average heart rate; BHdtot: total BH duration during training;
Bradycardia: average of ratio between HRstart and
HRmin during BH; CV: coefficient of variation. Correlation
between equation training load and reference value *: p<0.05; **:
p<0.01.
No differences were found for the phonemic fluency test score, the categorical verbal
fluency test score and the PS index score between before and after training. The IM
index was higher after training compared to before training (p<0.01). No
difference was found between the 40 controls participants and the 18 BHDs for the
PS
index and IM index. All Z-score are within±2 SD point of the norm [18]. The results of categorical verbal
fluency performed after training were negatively correlated with the total BH
duration during training (r=− 0.486, p=0.041) and the total number of BHs (r=− 0541,
p=0.020). There was a negative correlation between the years of BH practice and the
result of the phonemic fluency test before (r=− 0.590, p=0.010) ([Fig. 3a]) and after training (r=− 0.505,
p=0.033) without effect of age. The BH training loads assessed with eq. 1 were
correlated negatively with PS index after training (r=− 0.608, p=0.007) ([Fig. 3b]).
Fig. 3
(a) Correlation between the years of BH practice and results of
phonemic fluency test before training (r=− 0.590, p=0.010). (b)
Correlation between training load equation 1 and results of PS index after
training (r=− 0.608, p=0.007).
Discussion
This pilot study is a first attempt to establish a training load index capable of
estimating the stress induced by a BH training, while simultaneously studying its
possible impact on the BHDs memory performances in ecological situation. The main
results suggest that aTRIMP is the most representative measure of training load for
our well-trained BHDs. Training conducted under ecological conditions does not seem
to affect negatively the BHDs’ memory. However, the negative correlations observed
between years of BH practice and the phonemic fluency test indicate that long-term
exposure could have a negative impact.
Although many studies have focused on monitoring training load in activities such
as
endurance sport [22], limited research
has specifically focused on BHD training. The aTRIMP takes into account the BHD
internal and external loads. The top portion of the aTRIMP (HRstart)
considers the HR at the beginning of the BH. The higher the HR at start, the greater
the level of difficulty. Meanwhile, the denominator of the aTRIMP
(%HRmin) assesses the extent of the BH [8] as shown by the correlation between
HRmin during BH and total BH duration during training (r=− 0.610,
p<0.01). For BHs that last a long time, the lowest HRs are often associated with
the highest levels of hypoxia [23]. This
suggests that internal training load, represented by HRmin, can serve as
an index of both the intensity (state of hypoxia) and volume of stress experienced
during BH training. In our study, the BHDs, on average, allocated around 80% (±11%)
of the duration of their training to zone 3. This intensity zone, calculated by the
aTRIMP, therefore indicates that their training was more of a “moderate” type and
probably moderately hypoxic. Therefore, training at higher levels (aTRIMP>zone 3)
would allow for a hypoxic dose high enough to generate physiological adaptations
and, ultimately, improve the BHDs performance [24]
[25]. It has been shown
that the optimum intensity for maximizing the diving response and achieving longer
BH durations appears to be at a speed corresponding to 30% of the maximum oxygen
consumption [15]. Without speed
instructions, it is probable that the BHDs unconsciously adjusts their speed to this
intensity [26]. Our results also
indicate that there was a positive impact on IM, which was higher after the BH
training than before. This result is probably due to a learning effect, re-encoding
during the second session (after training) may increase performance. We also
identified a negative correlation between the BHdtot and performance on
the categorical verbal fluency test, specifically after the training suggesting that
more experienced BHDs’ prolonged practice could lead in the long term to alterations
in the neural networks involved in semantic memory processes. As with the effects
of
exercise, it is likely that a certain hypoxic dose (duration x frequency x oxygen
fraction) must be reached for observable changes in memory to occur. This dose may
vary according to individual factors such as training history and physiological
adaptation to hypoxia [27]. It appears
that the duration and intensity of exercise, the specific type of cognitive
performance assessed, as well as the physical condition of the participants, play
significant roles as moderators in determining the effects on cognitive function
[28]
[29]. A decrease in executive function in
elite BHDs compared to non-BHDs has been demonstrated, indicating that the duration
of BH training was related to this decrease [12]. Our study revealed no negative effect on memory performance in the
BHDs compared to the control group prior to BH training or when comparing tests
conducted before and after BH training. This suggests that a single BH training of
this intensity (zone 3 aTRIMP, 20 minutes of cumulative BH, at a speed of 0.8 m/s
for 33 BHs) for BHDs with five years of experience does not seem to influence the
memory. This may seem reassuring since this BHD training was a normal and usual
training. The utilization of aTRIMP zones within training protocols will require
future validation through the integration of muscle oxygenation measurements in
addition to cardiac and performance assessments. It is also important to recognize
that the other equations provided acceptable estimates of BH training load, despite
some limitations. Equation 1 did not take into account individual variability in HR
response during the BHD training, as it only considered the HRavg over
the entire training. Equation 2 did not incorporate the internal load factor,
neglecting the impact of BH’s response to external load demands. Equation 3
incorporates the individual's subjective perception using the RPE, but requires
prior familiarization with the RPE scale in order to use it more accurately [30] and does not take into account the
different parts of the BHD training, alternating repetitive BHs with more or less
long recovery periods. Moreover, it was observed that both the RPE and the RPD did
not exhibit any significant correlation with equations 1 and 2, as well as aTRIMP.
These findings suggest that these particular indices may not be well suited for
evaluating the physiological demands of BHDs’ training. In future, it will be
necessary to validate the aTRIMP and also test all the other training load indices
during a training follow-up lasting several weeks, while measuring oxygenation
parameters. Despite these limitations, our pilot study and the aTRIMP, in
particular, offers valuable information for the creation of a training load equation
specific to BHD, enabling coaches to manage their athletes’ training load.
Conclusion
Our pilot study indicates that of the different equations tested, the aTRIMP appears
to be the most representative for evaluating BH training. This type of training does
not adversely affect short-term memory performance. These results contribute to our
understanding of the evaluation of the BH training load and underline the importance
of considering the BH duration and intensity from an ecological point of view. A
longitudinal study with training follow-up will optimize the reliability and
reproducibility of these load indices.
