Investigating the relation between training intensity and acute HRV changes in endurance and power sports: an analysis of HRV4Training users data

Blog post by Marco Altini.

In quite a few previous posts (see the blog index on the right) I've covered acute HRV changes, i.e. changes in HRV following training. I've shown how we can use differences in HRV measurements between training day and the day following training to determine the impact of our training session on our physiology. What we typically expect in this situation is a reduced HRV on days following intense aerobic trainings. This method is one of the most common and reliable ways to use HRV, and is the principle behind the daily HRV advice provided by apps like HRV4Training.

Additionally, HRV4Training provides an analysis of acute HRV changes over a longer period of time (90 days), to determine more systematically what is the impact of training on HRV in your specific case. This analysis does not tell us much about the big picture or our overall physical condition, but it tells us something about how we recover from training of different intensities and can be helpful in making small daily adjustments in our training plan. See the image on the right for an example of the acute HRV changes analysis in HRV4Training.

In the last two blog posts I've shown 1) how training intensity reflects on HRV (rMSSD) on a dataset of 400 HRV4Training users 2) how such acute HRV differences are less obvious when we are not very consistent with our measurements, highlighting once more the importance of measurements consistency.

In this post I'll explore another interesting point when it comes to training and acute HRV changes, i.e. can we rely on this measure for different sports? Is monitoring day to day changes in HRV useful only for aerobic sports (e.g. endurance running) or can we benefit from this type of monitoring for other sports as well?

Before we start

A few points are worth mentioning before we start, to avoid confusion:

• Regardless of the sport you do, monitoring your physiology (HR and HRV) with respect to all sort of other factors (e.g. the different tags you can annotate in HRV4Training right after the measurement) can be very useful in determining correlates of physiological stress, and help you tweak your lifestyle and trainings (seethis post for analysis of HRV data outside of the training context). 
• Users performing different sports are different people and therefore their physiological response to training might change based on factors other than sport type. This being said I will consider only big subgroups (e.g. endurance and power sports) to limit this problem. 

What do we know from literature?

It is clear from literature that acute HRV changes reflect training load and training intensity and can be used to adjust trainings [1-4]. This is the whole principle behind HRV monitoring and HRV apps, and has been shown multiple times in various sports, mainly aerobic ones, for example running [2, 3], biking, triathlon [4], soccer, etc.

It is less clear if the same principle holds for other sports. This is an active area of research, and I'd suggest checking out Andrew Flatt's blog and research. Andrew is one of the most active researchers around HRV and training monitoring in athletes these days.

An interesting dataset was analyzed in this context by Berkoff et al [5], where American track-and-field athletes were split into two groups, endurance and power. The distinction was based on what the authors felt the predominant focus of the athletes’ training was. All running events, middle distance and longer, comprised the aerobic group, with all of the remaining events making up the power group (decathlon, heptathlon, the sprints that were 400 m or less in length, all hurdle events, shotput, discus, javelin, and hammer).

However, the authors did not have longitudinal data on HRV and training over periods of weeks to months, and therefore they could only look at the data cross-sectionally, and determine that there was no difference between groups in HRV values (i.e. all athletes showed rather high HRV values, as expected, regardless of the focus of their training). With only one data point per person, and without monitoring training, the authors couldn't determine the effect of different training types on acute HRV changes.

​As I try to stress often, apps like HRV4Training, ithlete or Elite HRV open great opportunities to better understand complex relationships between physiology, training and lifestyle factors, by outsourcing data collection to thousand of users and collecting reference points around the physiological measurements (e.g. tags in HRV4Training). Thus, we can try to answer our initial question using HRV4Training data, i.e. is monitoring day to day changes in HRV useful only for endurance sports (e.g. running) or can we benefit from this type of monitoring for other sports as well? 

Dataset

For this analysis, I used the last 3 months (October-December 2015) of measurements collected from HRV4Training users. I filtered out all users with less than 20 measurements and 10 trainings annotated during the selected time period, to make sure I had enough data to make meaningful considerations. This procedure left me with data from 597 users, summing up to 39387 HRV measurements.

Then, I split users into two groups, similarly to [5]:

• Endurance. User-selected sports: running, biking
• Power. User-selected sports: crossfit, body building, martial arts, powerlifting

The information I have is not extremely detailed, for example it is true that running could mean running 100m on the track and therefore not really aerobic, but among my assumptions are that most runners are endurance runners. Selecting only users that belong to the categories above left me with a total of 350 (239 endurance + 111 power) users and 23733 measurements.

The plots below show the distribution of anthropometrics data for the two groups:

​I preferred to use density plots instead of histograms due to the different sample size of the two groups. This way we can better see the differences in values without getting confused by the differences in sample size.

We can easily spot some expected differences between the endurance and power categories. For example, body weight is higher in power sports, same for BMI. On the other hand height is similar and endurance sports seem to be preferred by a broader set of people, typically of older age. All points that make sense. 

Finally looking at HR and HRV (rMSSD) values for the two groups we can see an almost perfect overlap in rMSSD values, together with a consistent reduction in HR for endurance sports (also to be expected given the different type of training):

The slightly lower rMSSD for endurance athletes might be due to an average higher age. However, age does not seem to influence much HR at rest, and the differences in HR shown above might be more related to the effect of endurance training on the heart.

Analysis of acute HRV changes

Acute HRV changes were computed as follows; I first determined day to day differences in HRV, and then contextualized the HRV change with respect to training. Then, for each user I averaged HRV changes on days following training, and on days not following training, to determine if there was any difference. 

In other terms, the questions we try to answer are the following; can we spot reductions in HRV on days following training? If so, can we spot greater reductions in HRV on days following more intense trainings? Finally, can we spot reductions in HRV in both endurance and power sports?

Similarly to what I was showing in the previous post on acute HRV changes, I computed HRV changes for different conditions:

• Rest day vs training day. Here we expect to see reductions in HRV on days following training.
• Training intensity. We expect to see greater reductions in HRV on days following more intense trainings.

From the plots below we can see that for endurance sports the results are as expected, with reductions in rMSSD on days following training, and greater reductions for more intense trainings, similarly to what was reported multiple times in literature as well as in previous analysis on this blog. We can see how rest days typically trigger a positive change in HRV, i.e. HRV is higher on the day following a day of rest. We can also see how easy trainings are also resulting in an increase in HRV the following day. Easy trainings (e.g. a recovery run) can have a stimulatory effect on parasympathetic activity, therefore explaining the higher rMSSD value. However, for both average and intense trainings we have clear reductions, i.e. negative HRV changes. The reduction in HRV is more marked for intense trainings, highlighting how we can use reductions in HRV as a marker of training load in endurance sports.

On the other hand, for power sports the relation is less easy to interpret. We can see for example how intense trainings can still be discriminated, since they consistently reduce HRV on the following day. However, the same does not apply to moderate intensity trainings.

​

Summary

​In this post I analyzed the relation between acute changes in HRV (rMSSD) and training intensity, on 350 HRV4Training users and 23733 measurements split between endurance and power sports. By looking at acute changes longitudinally over 3 months, we can analyze systematically what is the effect of subjectively rated training intensity on HRV.

​The data shows that trainings of higher intensity result in a reduction in HRV the following day, for endurance sports. These results highlight once again how HRV can be used as a valuable tool to measure training load and recovery in endurance sports (you can do the analysis yourself using your own data after collecting enough measurements and training annotations in the HRV4Training app, under Menu - Insights - Acute HRV Changes). However, for power sports we could find a reduction in HRV only after intense trainings. 

This analysis is another example of how we can use user generated data to try to answer some additional questions on the complex relationships between physiological data, training and lifestyle. Data were acquired in totally uncontrolled settings using mostly 60 seconds PPG measurements and subjective (self-reported) annotations. There might be confounding factors that I have not considered and therefore it will be interesting to extend this analysis later on including even more data and parameters. 
​

Follow @marco_alt

​Like the app and the blog?

If you like the app and or the blog, take a minute to review HRV4Training on the Apple store. ​

HRV4Training on Facebook

HRV4Training is on Facebook, I plan to use this page as a centralized place for feedback, discussions, feature requests, bug reports and to introduce new features and changes in the app. 

Especially for the ones that are not much into Twitter, feel free to use the FacebookHRV4Training page as a place to open discussions around HRV, training, performance and more. I'll try to be responsive so that the whole community can benefit

References

[1] Kiviniemi, Antti M., et al. "Endurance training guided individually by daily heart rate variability measurements." European journal of applied physiology 101.6 (2007): 743-751.
[2] Pichot, V., Roche, F., Gaspoz, J.M., Enjolras, F., Antoniadis, A., Minini, P., Costes, F., Busso, T.H.I.E.R.R.Y., Lacour, J.R. and Barthelemy, J.C., 2000. Relation between heart rate variability and training load in middle-distance runners. Medicine and science in sports and exercise, 32(10), pp.1729-1736.
[3] Buchheit, M., Chivot, A., Parouty, J., Mercier, D., Al Haddad, H., Laursen, P.B. and Ahmaidi, S., 2010. Monitoring endurance running performance using cardiac parasympathetic function. European journal of applied physiology, 108(6), pp.1153-1167.
[4] Plews, D.J., Laursen, P.B., Kilding, A.E. and Buchheit, M., 2012. Heart rate variability in elite triathletes, is variation in variability the key to effective training? A case comparison. European journal of applied physiology,112(11), pp.3729-3741.
[5] Berkoff, D.J., Cairns, C.B., Sanchez, L.D. and MOORMAN III, C.T., 2007. Heart rate variability in elite American track-and-field athletes. The Journal of Strength & Conditioning Research, 21(1), pp.227-231.