The case of Professor Sir Roy Meadow demonstrates that child protection is now the speciality that dare not speak its name. As the paediatrician answers a charge of serious professional misconduct at the General Medical Council for giving "seriously misleading and flawed" statistical evidence at Sally Clark's 1999 trial, he has been much vilified in past weeks - some websites are even nonsensically proclaiming he is another Harold Shipman. Surely no one can now be wondering why there are fewer and fewer doctors wanting to go anywhere near child protection.

Let's get something clear: there really are mothers and fathers and guardians who abuse their children. There are video evidence and guilty pleas to prove it. Indeed, as any police officer will tell you, you are most likely to be murdered by someone you know. Victoria Climbie was a horrific example of what happens when the possibility of abuse isn't fully investigated.

I haven't yet met a healthcare professional who enjoys reporting suspected child abuse to social services or the police: it weighs heavy on the mind. And yet the alternative - doing nothing - has consequences that can be even more appalling. The people I have met who work in child protection have been committed individuals who find themselves having to change their phone number regularly because of threatening calls.

The bottom line is that there is no single reliable test that will prove child abuse, which is why, in the end, it is a matter for the courts. Society as a whole has to decide what level of "safe" or "unsafe" evidence is going to protect parents from wrongful conviction, or children from an abusive guardian. Absolute certainty about such matters rarely exists.

· Every so often a clinical research paper comes along that isn't just academically significant, but practical. One such beauty - about childhood eye infections - has just been published.

Conjunctivitis is common. One in eight children have an episode of it every year. With a million episodes of conjunctivitis every year, it's the kind of thing you would expect to attract a vast amount of research, wouldn't you?

It seems not. The answer from Oxford, published recently in the Lancet, is that the standard treatment - a course of antibiotic eye drops - is actually no better than a placebo. Aha, but could that be because some of the infections are viral, rather than bacterial? Wouldn't this explain why the antibiotic drops didn't always work? The researchers thought of that. They swabbed the eyes - and even when a bacterial eye infection was proven, the difference in cure rates between the placebo and the antibiotic did not differ "significantly".

This is great news. We know that overuse of antibiotics leads to resistant strains - the headlines constantly remind us of it. So why not avoid using an antibiotic if it's not needed - say, unless the infection isn't settling down after a week or so. Why don't we let the body use its own natural immune response to clear up the infection for itself?

But apart from the cultural shift - we have grown to expect some kind of drops for our children's sticky eyes - we also have to consider the Department of Health's Medicines and Healthcare Products Regulatory Agency (MHRA). It recently decided that the antibiotic eye drops - chlor-amphenicol - are to lose prescription-only status and will now be on sale over the pharmacy counter. What a shame they are no more effective than a placebo.

· The writer is a GP in Glasgow