Considering our study divers following standard sports diving profiles not exceeding moderate exhaustion, no omitted safety stops and almost no decompression stops required, the observed percentage of divers with detectable bubbles in reference echocardiography was high and comparable to similar research . Even before repetitive dives on the same day and despite washout during the surface interval, bubbles were still detected on a regular basis but of a lower bubble grade when reentering the water compared to after ascending from the previous dive. However, apart from fatigue, no symptoms were recorded that could be related to a decompression incident.
Learning curve for audio Doppler self-assessment
We found that audio Doppler self-assessment can be learned to detect a reliable venous signal in the subclavian position within 1 minute after 2 attempts and within 30 seconds after 13 attempts. The self-assessment at the precordial position seemed to be more difficult and needed at least 5 attempts to detect a reliable blood flow signal within 1 minute and at least 18 attempts to manage below 30 seconds. This was mainly due to the prominent cardiac signals at this position. When self-assessment was done after the dives, more training of 2–4 attempts was needed to detect the same reliable signal within the mentioned time frame, probably due to the effects of lifting heavy diving gear while getting undressed and therefore limited fine motor skills at this moment. A limitation to generalize our results might be that we used an 8 MHz probe instead of the more commonly used 4 MHz probe for such assessments. This might influence results especially in heavier divers. Further, two tasks were learned within the same learning curve: reliably finding of a venous audio Doppler signal and recognizing bubble signals within the venous signal. This could potentially lead to missing of more bubble signals at the beginning of the learning curve and compared to the end, an effect that could lead to better results than described here. However, considering that basic skills of anesthesia and acute medicine require 50–60 trials to achieve an 80–90% success rate  and are a bit more complex than the skill taught here, the number of attempts needed to reliably hear and interpret a venous Doppler signal is comparably low. The same learning curve as in our study was found in beginners when performing a simple virtual surgical task .
Measurements from the precordial position showed around one-third higher occurrences of detectable bubbles in both audio self-detection and experienced examiner audio-detection, probably due to the precordial position also including blood flow from the lower part of the body with potentially higher inert gas flow due to fin swimming. This must not necessarily be contrary to previous findings of better bubble detection at the subclavian site , since it is easier to establish a reliable subclavian Doppler recording as also seen in our study. Further, in the precordial position, we have seen a higher amount of especially higher bubble grades. In the event that bubbles were detected by audio Doppler, the bubble grade was typically lower compared to the echocardiographic exam despite no significant time lag. The average difference in our study was 2 bubble grades lower in audio Doppler detection. Considering the number of measurements in our study, this difference can be relevant in the prediction of decompression outcomes . The fact that the highest bubble grade 4 in the audio Spencer scale is not really comparable to the more differentiated high bubble grades 4 and 5 in the Eftedal-Brubakk echo scale does not create any shortcomings in our results since our divers were not able to grade correctly. In general, a reliable bubble grading using audio Doppler assessment was not possible for lay-people and for both, lay-people and experienced ultrasonographers, underestimated compared to echocardiographic results. Especially lacking noise reduction might have influenced audio Doppler grading, but this is an important aspect of the usability of this method in dive site reality. In summary, we conclude that divers are able to learn audio Doppler self-assessment with a standardized training and a reasonable number of trials, but cannot determine their bubble grade and decompression risk reliably. However, in being able to recognize bubbling at all and especially when they have high bubble grades, they are able to generate relevant qualitative information on major bubbling only for potential decision making, e.g. on safer surface intervals, in order to reduce decompression stress. Especially when they plan to reenter the water, audio-detectable bubbling from the previous dive can lead to advisable stretching of surface intervals. Specificity in general and sensitivity for higher bubble grades proved to be acceptably high to provide qualitative information on decompression stress in order to indicate measures for avoiding decompression incidents.
After a gap of six months without any training, the previously achieved skill level was reliably preserved: a venous Doppler signal of good quality was self-detected within the same time compared to the end of learning curve after initial teaching and individual anatomical ultrasound windows were remembered. This long-term skill retention of a similar condensed 45-minute standardized training has already been shown in a study on teaching central line placement . Hence, the training proved to be suitable enough to generate a stable practical skill level over time.