Well this doesn't sound like some overbearing Orwellian surveillance nightmare at all does it? Perhaps not the best choice of wording by the Safety Gods who decide these things. Health surveillance is basically 'hearing tests', or 'audiometry' and Regulation 9 of the Noise Regs says that you needs to have a programme of health surveillance for anyone who remains at risk of noise while at work. IE anyone who is still exposed to over 85dB(A) or 137dB(C), ignoring the reductions offered by the hearing protection.
Why do it?
Shhh, say it quietly, but the human ear is great at keeping glasses on your face but are not all that brilliant at their main job of hearing and are particularly rubbish at detecting small reductions in hearing ability over time. Health surveillance is therefore the back-stop to all the other controls and provides a means of identifying any ongoing reductions in hearing ability, usually long before the person themselves become aware of it. It acts as a double-check that the reductions measures in place are working and allows changes to be made if people are showing signs of still being at risk.
It also allows people for whom noise is more of a risk than other to be identified who can then maybe have additional protective measures. For example, a small drop of hearing in a woman in her 20s would probably mean she still has excellent bat-like hearing, while a small drop in the hearing ability of someone who already has problems could be the difference between being able to hear normal conversation or not. Clearly that second person is more at risk than the former as the impact on them is much greater and the health surveillance can help identify them. Don't forget, it is true in health and safety in general that you cannot assume everyone is the same and treat them equally, some people may be more at risk in some situations than others and you need to account for this at work, and this applies just the same to noise risks.
What is health surveillance?
At it's simplest, it is a basic hearing test to determine how good their hearing is and compare it to the standards which are expected for their age. So the old 'sit in a box and press a button when you hear a beep' game. It's an ongoing process where over time you begin to be able to compare people's results not only against standards for their age but also against their own historic data meaning you can identify small changes over time, hopefully well before they begin to notice it.
Should you wish to dive into this more deeply then have a gander at the Health Stuff section.