What’s your resting heart rate?
A simple question with a not-so-simple answer.
When tracking data, a few things matter more than others:
accuracy of the sensor
consistency and timing of the protocol, allowing for a proper physiological interpretation of the data
For example, we need to use an accurate sensor to measure resting heart rate, a sensor that is actually able to capture this information correctly. The good news here is that the most common wearables out there can all measure heart rate correctly in the absence of movement.
Secondly, we need a consistent protocol. Why is this important? When we measure resting heart rate or HRV, what we want to do is to capture relative changes over time in relation to changes in e.g. physiological stress or cardiorespiratory fitness.
If our protocol is inconsistent, even if the sensor is accurate, the data will be meaningless. For example, an Apple Watch, which is sampling randomly during the day and night, is the last device I would use to track my physiology. The context of the measurement changes all the time: I might be doing different things, and therefore the data will reflect what I am doing, not what my response to what I do is, this is the key point to understand here.
A consistent protocol means for example measuring every day after waking up, or using the full night of data during sleep, basically things we can do every day in similar ways, what we call a reproducible context.
Once we have an accurate sensor and a consistent protocol, the last step to consider is the timing of the protocol / measurement. The timing matters because of the physiological interpretation of the data. For example, if we are interested in tracking resting heart rate (and / or HRV), it is key to measure these parameters far from stressors: we want to capture our response to stress, not our behavior. If we measure too close to stressors, our physiology is simply reflecting the stressor itself (e.g. a low HRV or high heart rate right after exercise or after a meal). We already know our behavior, the question to answer is: how did we respond to that behavior? Did it impact our physiology negatively or not? The only way to know is to measure in a reproducible context, reasonably far from stressors and confounding factors.
Let’s see how common wearables perform when it comes to measuring your resting heart rate. Well, to start with, resting heart rate is reported differently by most wearables:
Oura: your lowest heart rate is called resting heart rate.
Garmin: your lowest 30-minute heart rate in the last 24 hours is called resting heart rate.
Whoop: your night average, weighted more strongly during deep sleep, is called resting heart rate.
Apple Watch: random samples of your heart rate when not moving, possibly excluding data collected during the night, is called resting heart rate.
Oura also reports the average of the night, and that is what I highly recommend using as resting heart rate if you prefer not to measure in the morning for some reason.
Using the lowest heart rate is artifact-prone (any min or max could be very far from the truth in case of inaccuracies in the measurement, which are frequent for optical sensors) and also compares data collected at different times every day, (your lowest could be at 2 am one day, and at 7 am another day, confounding the data with your circadian rhythm and sleep stages) hence I think it is best to use the average. When using the minimum, there is no consistency in the protocol, as the timing changes.
Whoop's approach is in fact extremely similar to Oura's night average, the data gets a bit more noisy with their weighting (because of course, deep sleep cannot be detected accurately), especially for HRV, but for resting heart rate, it is basically equivalent to using just the average of the night. Note that also here the consistency of the protocol is flawed, because depending on how much deep sleep you get (well, not really, let’s say depending on how much deep sleep the device thinks you got), your resting heart rate changes, and this can change every night.
Garmin's approach is quite problematic as sometimes those 30 minutes are even during the day or at dramatically different times, making the data of no use. Using 30 minutes is better than using a single minimum, but the issue here is that 24 hours of data are used (if this is still the case, it could also be that the algorithm has changed, making past data inconsistent, another “feature” of wearables). In practice, issues here are rare as under most circumstances your lowest 30 minutes will be in the night, but still, it is an inaccurate assessment, and also prone to the issue above of inconsistent timing.
Please see my blog here where I looked at Oura, Whoop and Garmin longitudinally for 3 months and in relation to ECG data.
A final note on Fitbit (or the Apple Watch, which seems to behave similarly, or maybe even worse, actually discarding night data). Fitbit implements a poor approach, using the average heart rate of "resting" periods, defined as non-movement periods, even during the day. Seems like the engineers who came up with that algorithm many years ago forgot to talk to any physiologists. This data reflects your behavior (e.g. you are more active, your resting heart rate is higher) more than your physiological response (which needs to be measured at rest, i.e. far from stressors).
Keep in mind that a morning measurement allows you to capture data as far as possible from the previous day's stressors, as the morning comes after the night. This is ideal in terms of interpreting the data and doing something with it.
A morning measurement is what I would call your resting heart rate: it reflects your response to stressors, and your "recovery" (for lack of a better word), as opposed to just your behavior (which is what happens when you measure too close to stressors).
The absolute value of your resting heart rate will depend also on the protocol (e.g. if you are lying down or sitting, it will be different), but as you know already, what matters are relative changes over time, not absolute values. However, the point here is that without consistency of the protocol and timing, even relative changes over time will be meaningless, as we lose the physiological meaning of the measurement.
For these reasons, if you really do not want to measure your physiology first thing in the morning, intentionally, when you are in control of the data and outcome, a night average is my second-best option.
Take it easy.
Marco holds a PhD cum laude in applied machine learning, a M.Sc. cum laude in computer science engineering, and a M.Sc. cum laude in human movement sciences and high-performance coaching.
He has published more than 50 papers and patents at the intersection between physiology, health, technology, and human performance.
He is co-founder of HRV4Training, advisor at Oura, guest lecturer at VU Amsterdam, and editor for IEEE Pervasive Computing Magazine. He loves running.
Social:
Twitter: @altini_marco.
Personal Substack.